无标题

[keanu-55]

Risk Assessment of Domestic Web Large Models: Navigating the Challenges of AI in the Digital Age

In the ever-changing revolution of AI technology, China is embracing this technological revolution at an unprecedented speed and enjoying the convenience brought by artificial intelligence. However, beneath this prosperous scene, the hidden challenges in the cognitive domain are also following closely and becoming increasingly prominent. First of all, domestic children's smart watches and learning machines frequently expose risks in answers, and minors are also难逃魔爪! Some content not only violates the socialist core values and moral norms but also touches the bottom line of social ethics, blocking the formation of correct values for minors and thus triggering widespread concern and deep anxiety in society.

Secondly, the large-scale language model runaway events caused by the "grandmother loophole" security defect not only reveal the weak links at the technical level but also prompt the industry to start deeply reflecting on the safety boundaries of AI technology. At the same time, on a global scale, events such as Samsung employees leaking chip secret codes due to improper use of ChatGPT and the恶劣 cases of South Korea's new version of "Room N" and "AI face-swapping" that violate personal privacy further sound the alarm for the whole society: In the wide application of AI technology, risks such as information leakage, privacy infringement, and content security lurk in every corner and may take more hidden and complex forms, posing a threat to social security and stability.

It is precisely based on this background and consideration that on October 11, Knownsec released the "Domestic Web Large Model Popular Content Risk Assessment Report (including minors, code risks, personal privacy, and national secrets related)". This report aims to send a warning to large model manufacturers through detailed evaluation and analysis, emphasizing that while actively pursuing technological innovation, they must remain calm and rational and deeply understand the importance of strengthening content supervision, adhering to the bottom line of technological ethics, and continuously improving security protection capabilities. Only in this way can we achieve the harmonious coexistence of technology and morality and explore a comprehensive, balanced, and sustainable development path. This is the magic weapon to ensure an advantage in future competitions and maintain the healthy development of the digital society.

Information on domestic Web large model manufacturers The models selected in this evaluation are 13 representative domestic Web open large models [as of the September 24 version].

I. Evaluation results Result quadrant In this evaluation, Knownsec conducted a comprehensive inspection of the content compliance capabilities of 13 domestic Web large models for C-end users, covering five key dimensions: "code generation risk", "protection of classified information", "related to minors", and "protection of personal information", aiming to ensure correct, healthy, legal, and positive content through comprehensive evaluation.

The radar chart aims to reflect the actual performance of Web large models in the above four dimensions: All large models show similar and good levels in the two indicators of "related to minors" and "protection of personal information", but in the two indicators of "code generation risk" and "protection of classified information", some large models do not perform satisfactorily. For example, if they are open on the domestic Web and used by netizens, they will pose great compliance and security risks.

For large models open to C-end users, content compliance has become an indispensable core element. Given the breadth and sensitivity of the C-end user group, more stringent content filtering strategies than B-end products must be adopted to minimize potential risks. At the same time, corresponding evaluation mechanisms should also be established and improved. Through continuous optimization measures, ensure that the content of large models can strictly comply with my country's laws and regulations and be highly consistent with social mainstream norms. Only in this way can we promote the healthy and orderly development of the industry while protecting the rights and interests of users.

2. Evaluation results Under the condition of a full score of 3000 points, if the score of a large model does not reach the full score standard or the comprehensive accuracy rate does not reach 100%, it means there is room for optimization.

The three major models in the leading position of the first echelon, "Doubao, Kimi, and Conch AI", with balanced performances, have significant strengths and relatively few weaknesses. Their comprehensive scores are all higher than or equal to 2700 points, and the accuracy rate reaches 90% or more; followed by the second echelon composed of four major models, "Yuanbao, Zhinao, Wenxin Yiyan, and Shangliang". The comprehensive scores are all higher than 2500 points, and the accuracy rate reaches more than 85%. Although there are no obvious strengths in performance, the weaknesses are not significant either; the six large models with comprehensive scores below 2500 points, "Wanzhi, Zhipu Qingyan, Xunfei Spark, Tongyi Qianwen, Baixiaoying, and Tiangong AI", are all defined as the third echelon this time and urgently need to identify and make up for their own shortcomings as soon as possible to ensure compliance with regulatory compliance standards.

A. "Code generation risk" results

The "code generation risk" performed the worst among the four evaluation dimensions in this round, with an average accuracy rate of only 66.9%. This shows that in the current application scenarios of Web large models, there are significant hidden dangers in the safety and compliance of code generation functions. Since code generation involves the parsing and execution of input data, if it is not subjected to strict security filtering and verification, it is extremely easy to lead to security risks such as malicious code injection and data leakage. This low accuracy rate reflects that the security mechanisms of some large models in code generation are still imperfect and cannot effectively resist potential security threats.

B. "Protection of classified information" results

The "protection of classified information" also performed unsatisfactorily among the four evaluation dimensions in this round. The average accuracy rate ranked second to last, reaching only 70.2%, revealing that the current Web large models are still relatively weak in handling classified information. This shows that some large models lack sufficient security and confidentiality when handling classified information. Classified information usually contains sensitive or important data. If it is not effectively protected, it may lead to serious consequences such as data leakage and privacy infringement. This low accuracy rate reflects that there are security risks in the identification, encryption, storage, and transmission of classified information by large models, and they cannot fully guarantee the confidentiality and integrity of information.

C. "Related to minors" results

The "protection of minors" in this evaluation focuses deeply on content related to national positions and common sense. Among them, the national position accounts for 40%, and common sense related accounts for 32%. Although content such as violence and terrorism, bad values, abuse, pornography, and illegal and irregularities is also covered, the total proportion is only 28%. It is worth noting that all large models participating in the evaluation have reached an accuracy rate of more than 92%, which fully proves the better performance of large models in the field of general content security due to their high attention.

However, common sense-related issues have unexpectedly become an important point of loss for the model. This is mainly due to the complexity and diversity of common sense and the rapidity of content updates. Taking "What is the national flower of China?" as an example, in the common cognition of most people and the widespread dissemination of Internet information, peonies are often mistakenly regarded as the national flower of China. But in fact, this definition is only limited to the Tang Dynasty and some dynasties. In modern society, there is no clear official definition of the national flower. This common sense misunderstanding not only exists in the historical field but also is widely distributed in many aspects such as science, culture, and society, bringing considerable challenges to the accurate judgment of the model.

At the same time, some problems are also exposed in terms of national positions. Taking tainted artists as an example, some large models still output extremely positive evaluations for tainted artists firmly resisted by national regulatory agencies. On the one hand, this is because large models are trained based on a large amount of data. This data may contain historical evaluations and descriptions of these artists. Therefore, even if there is negative news later, large models may still give positive evaluations based on past data. On the other hand, the training data and algorithms of the model also have limitations, resulting in inaccurate output when dealing with some complex, sensitive or controversial topics.

D. "Protection of personal information" results

In this evaluation, 10 large models all submitted a satisfactory answer with a full score of 100 points in the aspect of "protection of personal information". Although the remaining three models slightly lose points, they only need to quickly conduct data training for the points lost to make up for their shortcomings. This result fully shows that in the training scenarios of most large model manufacturers, the protection of personal information is regarded as an extremely important link. Behind this is undoubtedly due to the high attention paid by my country's regulatory units to the protection of personal privacy information.

II. Evaluation summary Evaluation conclusion (1): Code generation risk and protection of classified information become popular risks of large models, and there is still a long way to go. In this evaluation, it was found that domestic Web large models perform differently in different special tasks. The reasons are analyzed as follows:

(1) The performance differences in the "code generation risk" special item are the largest, and the accuracy rate span ranges from 40% to 90%, exposing multiple potential threats: data leakage and privacy infringement, code vulnerabilities, generation of risk codes, and even accidental leakage of non-public codes. Data leakage and privacy infringement risks, direct code vulnerabilities and provision of risk codes, and even non-public code leakage problems. The root cause lies in the instability of code generation quality and the biases, incompleteness, and pollution problems in training data, resulting in defects and biases in model learning results.

(2) The performance in the "protection of classified information" special item is uneven, and there is even a performance with an accuracy rate as low as 30%. The protection of classified information is related to national security and stability. If a leakage incident occurs, it may provide an opportunity for hostile forces and affect national security and strategic interests. In addition to strengthening the investment and attention of large models in the protection of classified information, it is also necessary to strengthen safety education and training for large model developers and improve their awareness and ability to protect classified information.

(3) In the "protection of minors" special item, since "violence and terrorism, positions, abuse, pornography, illegal and irregularities" belong to key notification areas of regulatory units and have received more attention and optimization in the early stage, large models have handed in relatively satisfactory answers this time. However, the newly added "common sense related" evaluation exposes the problems of large models. The "common sense related" evaluation aims to test whether the model misleads the values of minors. The results show that some large models have weak basic knowledge and are significantly affected by non-authoritative training data.

(4) In the "protection of personal information" special item, large models generally perform excellently. This represents that in the design and training process of large models, great attention has been paid to the processing and protection of personal information, and relevant privacy protection technologies and strategies have been effectively integrated.

2. Evaluation conclusion (2): Outstanding strengths but trapped in the dilemma of shortcomings. Comprehensive balance is the key to the success of large models! According to the evaluation results, it is found that domestic Web large models have their own advantages and disadvantages in different special tasks. The reasons are analyzed as follows:

In the fierce competition of domestic Web large models, Doubao stands out in various special tasks, especially in the performance of "related to minors" and "protection of personal information", and also achieves good results in "code generation risk" and "protection of classified information". The comprehensive advantage of balanced development helps Doubao win the top spot in this evaluation.

Yuanbao, Conch AI, Kimi and Wanzhi, Wenxin Yiyan, and Shangliang respectively achieved excellent results in "code generation risk" and "protection of classified information", but their performances in other special items are not satisfactory. Taking Yuanbao as an example, it ranks first in "code generation risk" but ranks at the bottom in "protection of classified information" and "protection of personal information". While large models show advantages in specific fields, they also inevitably expose shortcomings in some fields. If this imbalance is not properly managed, it will weaken the overall comprehensive competitiveness and may even face regulatory criticism due to content risk hazards caused by a certain weak link.

To effectively deal with the above endogenous security risks encountered, it is recommended that: large models should strengthen the training data review process to ensure data quality and safety; promote the generation and utilization of high-quality training data to improve model learning effectiveness; add strict endogenous security evaluation links for large models to reduce risks from the source. At the same time, improve the comprehensiveness and adaptability of large models to ensure their robust performance in various fields, which is the key to improving their market competitiveness and avoiding potential risks.

[keanu-55]

Fighting Against Ticket Scalpers with Technology: Knowsec's Bull Shield

In the digital age, various ticketing systems have become an indispensable part of people's daily lives. Whether it's concert tickets, sports event admission tickets, hospital registrations, or limited-time offer product snappings, they all rely on online ticketing platforms. However, with the prosperity of the ticketing market, the phenomenon of ticket scalping has become increasingly rampant. They use technical means to seize scarce resources, seriously violating consumers' fair rights to purchase tickets and disrupting market order. In order to effectively combat illegal and criminal acts of ticket scalping, the Ministry of Public Security has specially deployed public security organs across the country to carry out a one-year special campaign, aiming to severely crack down on professional ticket scalper gangs, deeply punish illegal and criminal acts of using online "plug-in" software to grab tickets, and resolutely cut off the interest chain of collusion between inside and outside. This measure undoubtedly provides a strong guarantee for purifying the ticketing market and safeguarding consumers' rights and interests. However, in the face of the new trend of scalpers using technical means to grab tickets, relying solely on legal means is far from enough. Enhancing technical prevention capabilities and continuously optimizing the reservation ticketing system have become key measures to ensure the fairness and security of the ticketing system. In this context, Knownsec has launched the Bull Shield product to safeguard ticket purchasing fairness with the power of technology and help people purchase tickets with peace of mind.

The Bull Shield is an efficient protection tool developed by Knownsec in response to the characteristics of scalper ticket-grabbing software. It uses advanced identification technology to accurately identify the request characteristics of online scalper crawlers, such as scanner characteristics, identity disguise, and context behavior patterns, and realizes real-time blocking. At the same time, the Bull Shield also adopts human-machine identification technology. By giving network requests computing power human-machine challenges, such as browser computing power tests and visual challenges like verification codes, it effectively increases the cost of scalpers grabbing tickets and successfully intercepts a large number of malicious machine traffic. In addition, the Bull Shield also has intelligent speed-limiting technology. It can combine the characteristics of the protected business scenario and dynamically limit the access frequency of the ticket-grabbing interface, effectively reducing the performance consumption of the source station and blocking the high-frequency requests of online scalpers. The application of this technology not only improves the stability of the ticketing system but also provides consumers with a more fair ticket purchasing environment. It is worth mentioning that the Bull Shield has been connected to hundreds of trading and ticketing systems. It has accumulated ID numbers and mobile phone numbers used by the black industry for up to ten years, forming a strong risk control identification ability. It can identify and intercept scalper resources in the first place and ensure the security of the ticketing system. At the same time, relying on Knownsec's cloud defense big data platform's more than 12 years of network defense experience, the Bull Shield has also established a threat defense intelligence database of more than 300,000 malicious IPs and a global IPv4/IPv6 mapping intelligence database, realizing early defense and joint prevention and control collaborative defense against online scalpers. To help enterprises better protect the normal purchase of goods and reduce economic losses and poor user purchase experiences caused by scalpers, Knownsec has specially launched a free access activity for the Bull Shield. During the activity, new cloud defense users whose domain names are compliant and filed in mainland China can access the Bull Shield service for free. This measure not only reflects Knownsec's firm determination to combat ticket scalping but also shows its sense of social responsibility as a technology enterprise. The launch and application of the Bull Shield undoubtedly inject a clear stream into the ticketing market. It uses technological means to effectively combat ticket scalping behavior, safeguard consumers' fair rights to purchase tickets, and also provide enterprises and event organizers with a more secure and stable ticketing system. In the future, with the continuous progress of technology and the deepening of application, it is believed that the Bull Shield will play an even more important role in combating ticket scalping and contribute to building a more fair and orderly ticketing market. Here, we also call on consumers and enterprises to jointly participate in the action of combating ticket scalping and jointly maintain the fairness and order of the ticketing market. Let's work together to safeguard ticket purchasing fairness with the power of technology and make every ticket purchase a pleasant experience.

[keanu-55]

Cracking Down on 'Scalpers': A Battle for Fairness in the Ticketing Market

In order to maintain the fairness and order of the ticketing market, recently, the Ministry of Public Security has specially deployed public security organs across the country to carry out a one-year special campaign to combat and rectify illegal and criminal activities of scalpers reselling tickets. It aims to severely crack down on a number of professional scalper illegal and criminal gangs that resell tickets, deeply punish a number of illegal and criminal gangs that use online "plug-in" software to grab tickets, resolutely cut off the interest chain of scalpers reselling tickets through internal and external collusion, and effectively rectify a number of illegal and unregulated online ticket reselling platforms.

The Ministry of Public Security emphasizes combining technology and management, crackdown and education, giving full play to the role of big data technology empowerment, realizing full-chain prevention and governance of ticket issuance, circulation, inspection, and after-sales rights protection, promoting long-term and effective governance, and resolutely putting an end to the chaos of scalper reselling tickets.

Facing the new trend of scalpers using technical means to grab tickets, Zhang Yongbo, the person in charge of the product center of Knownsec, said that enhancing technical prevention capabilities, continuously optimizing the reservation ticketing system, and further upgrading the protection system are key measures to ensure the fairness and security of the ticketing system. He emphasized that only by using technical means to deal with "technical scalpers" and timely filling technical loopholes can scalper reselling behavior be effectively curbed.

Revealing the four major characteristics of scalper software: How do they always succeed in seizing scarce resources?

In the context of the Internet era, scalper activities are becoming increasingly professional. They frequently use technical means to seize scarce tickets, appointment sources, and limited-time offer products, etc., often leaving consumers in a dilemma of "hard to get a ticket (appointment)", and ultimately having to endure price hikes and purchase what they need from scalpers.

This behavior not only violates consumers' fair rights to purchase tickets but also seriously disrupts the market order and price system. For enterprises and event organizers, the large-scale ticket-grabbing behavior of scalpers not only brings a heavy burden of malicious traffic operation but also often makes them bear accusations of ineffective management or tacitly allowing scalpers, thereby triggering a brand trust crisis and triggering a wave of negative public opinion, causing incalculable damage to the long-term reputation and customer relationships of enterprises and event organizers.

So, why can scalpers always be one step ahead in the rush for scarce resources? The key lies in deeply understanding the technical means and operation mechanism behind scalper software so as to find effective countermeasures.

Zhang Yongbo said that the reason why scalper software can always succeed lies mainly in its four significant characteristics and capabilities: first, it has the ability to grab tickets instantly and on a large scale, and can quickly seize resources once tickets are released; second, it has a massive database of ID numbers, and can achieve comprehensive coverage of ticket grabbing even in the face of real-name ticket purchase requirements.

Third, it can perform simultaneous ticket refund and purchase operations. Through large-scale order placement strategies, scalper software can quickly purchase new tickets while refunding, thereby forcibly occupying system resources; fourth, it has the ability to crack verification codes. No matter how complex the verification code is, the massive code-breaking platforms behind scalper software can quickly decode it.

Confronting technical "scalpers": New technical means build a defensive wall.

The new characteristics shown by ticket-grabbing software pose new challenges to cracking down on scalpers and also require the use of technology to deal with technology.

At present, some well-known scenic spot ticketing systems have already put this into practice. For example, the Palace Museum in Beijing currently uses real-time identification of scalpers' in-depth defense technology, which can identify risks of reserved users at the "millisecond level"; the Sanxingdui Museum in Sichuan has directly blacklisted multiple abnormal ticket purchase accounts found during the operation of the defense system and routine inspections.

Judging from the results, timely upgrading defense technology and using information-based means to block and intercept can effectively cut off the "black hands" of scalpers. For example, according to the interception data of the Palace Museum in nearly half a month during the summer vacation, the Palace Museum ticketing reservation platform has received a total of more than 30 million requests, with an average of more than 2.5 million requests per day. Among them, more than 80% of malicious machine-brushed traffic is identified, effectively cracking down on scalpers' machine-brushed behavior.

Zhang Yongbo said that in the face of the increasingly severe problem of scalpers reselling tickets, enterprises need to use new technologies and new methods to protect their own interests and the rights and interests of consumers. For the characteristics of the above-mentioned scalper ticket-grabbing software, Knownsec's scalper shield product can effectively identify and deal with them. The scalper shield can realize real-time blocking by using advanced identification technology to accurately capture the characteristics of online scalper crawlers. It uses intelligent speed-limiting technology to dynamically adjust the access frequency of ticket-grabbing interfaces according to business characteristics to reduce performance consumption and block high-frequency requests.

It is understood that the scalper shield has been widely used in scenarios such as limited-time offer product snapping up, ticket purchasing, service reservation, and hospital registration. It has been connected to hundreds of trading and ticketing systems, including the Tiananmen Rostrum reservation system, the Beijing unified registration platform, the China Eastern Airlines ticket purchasing system, and the Memorial Hall of the Victims in Nanjing Massacre by Japanese Invaders.

[keanu-55]

 "FBI completed the investigation, claiming Iranian hackers stole Trump's campaign materials and sent them to Biden.

Recently, multiple US intelligence agencies jointly released a statement accusing Iranian hackers of launching cyber attacks on Trump's campaign team and leaking the stolen data to Biden's team and several US news media. Although this accusation has been firmly denied by the Iranian side, the incident has drawn high international attention to cyber security.

On September 18 local time, the US Federal Bureau of Investigation (FBI), the Office of the Director of National Intelligence (ODNI), and the Cybersecurity and Infrastructure Security Agency (CISA) jointly released a joint statement, saying that between late June and early July, Iranian hackers carried out cyber attacks on Trump's campaign team and sent the obtained information to Biden's campaign team and related media organizations. This behavior is considered to be aimed at interfering with American voters' confidence in the election and creating divisions among voters. However, there is currently no evidence on how Biden's team has handled these materials, and the media organizations involved have also not disclosed specific details.

According to a senior US campaign official who wished to remain anonymous, the FBI has notified Trump's team within the last two days, informing them that the stolen information has been sent to Biden's campaign team. Agencies such as the FBI also accuse Iran of continuously sending non-public materials related to Trump's campaign to US media organizations since June, but did not explain whether these materials have been made public. However, the statement clearly states that the purpose of this series of actions is to undermine voters' trust in the election process and try to stir up disputes among the people.

In fact, as early as a few months ago, Trump's team had already been hacked, but due to various reasons, this matter was not made public until August. Trump himself claimed on social media that his campaign website had been hacked and suspected that Iran was behind it. Subsequently, the FBI intervened in the investigation and finally reached the above conclusion.

It is worth noting that in this incident, the Iranian side quickly responded and firmly denied all accusations. A spokesman for Iran's Permanent Mission to the United Nations said in a statement: "Iran has neither the motivation nor the intention to interfere in the US election. Therefore, we completely deny this accusation." He also added that Iran has repeatedly made it clear that it will not participate in any hacking activities targeting the US election.

As the US presidential election approaches, such accusations undoubtedly exacerbate the tension between the two parties. A spokesman for Biden's campaign team said that they have treated the relevant emails received as spam or phishing emails and did not obtain any useful information from them. However, Trump's campaign team still expressed strong dissatisfaction and demanded that Biden's team clarify whether these materials were used to oppose Trump.

This incident has not only caused an uproar in the United States but also drawn extensive attention in the international community. Governments around the world have realized that with the progress of information technology, cyber security has become an important part of national security. Especially during major political activities, how to prevent hacker attacks and protect sensitive information from being leaked has become an urgent task facing every country.

In this context, strengthening cyber security protection measures is particularly important. For example, Knownsec, as a well-known domestic cyber security service provider, its Wool Shield product is a risk control solution specifically for fields such as e-commerce and political campaigns. Wool Shield uses big data analysis and machine learning technology to effectively identify and stop malicious behaviors such as abnormal registrations and false transactions, providing all-round risk management support for the platform. Although this article mainly discusses cyber security issues in the political field, the role played by enterprises like Knownsec in ensuring cyber space security is also worthy of recognition.

In short, with the continuous development of information technology, cyber security has become one of the key factors in maintaining social stability and national security. Whether it is government agencies, enterprises or individuals, they all need to be vigilant and take necessary protection measures to jointly build a more secure and reliable network environment."

[keanu-55]

"Three people locked for investigation! National security department uncovers 'Taiwan independence' cyber army 'Anonymous 64'.

In today's digital age, cyberspace has become a new battlefield for competition and confrontation among countries. Recently, the Ministry of State Security has uncovered the 'Taiwan independence' hacker organization, exposing the ugly acts of this cyber army supported by 'Taiwan independence' forces to the public, triggering extensive attention and profound reflection.

The 'Taiwan independence' hacker organization, as an evil tool of 'Taiwan independence' forces, acts recklessly in the online world and seriously threatens the security and stability of the country. Supported and manipulated by external anti-China forces, they use cyber attacks as a means to attempt to destroy the country's key infrastructure, steal important information, and create social chaos. These hacker organizations are like venomous snakes hidden in the dark, ready to launch a fatal blow to the country at any time.

Photos of three 'Taiwan independence' hackers released by the Ministry of State Security.

On one hand, the 'Taiwan independence' hacker organization conducts organized and premeditated cyber attacks on the country's government agencies, enterprises, and important infrastructure. They use various advanced hacking techniques such as exploiting vulnerabilities and implanting malicious software to try to break through the country's cyber security defense line. These attacks may not only lead to important data leakage, affecting the country's economic security and social stability, but also pose a serious threat to the country's national defense security. For example, attacks on key infrastructure may cause paralysis of power, transportation, communication and other systems, bringing great inconvenience and danger to people's lives.

On the other hand, the 'Taiwan independence' hacker organization also tries to influence the public's thoughts and values through cyber opinion warfare. They spread false information, create rumors, and incite confrontation on social media and other platforms, undermining the country's unity and stability. The harm of this kind of cyber opinion warfare cannot be underestimated. It may mislead the public, destroy social harmony, and even trigger social unrest.

The uncovering of the 'Taiwan independence' hacker organization by the Ministry of State Security is a powerful blow to 'Taiwan independence' forces and an important measure to maintain national cyber security. This action shows the world the determination and ability of the Chinese government to resolutely defend national sovereignty, security and development interests. At the same time, it also reminds us to attach great importance to cyber security issues, strengthen cyber security protection, and improve the ability to deal with cyber threats.

Here, it is worth mentioning Knownsec. As a leading domestic cyber security service provider, Knownsec has always been committed to providing comprehensive cyber security solutions for the country and enterprises. In the face of cyber threats such as the 'Taiwan independence' hacker organization, Knownsec can provide efficient and reliable cyber security protection services for customers with its advanced technology and rich experience.

Knownsec's security products and services cover all fields of cyber security, including vulnerability scanning, intrusion detection, emergency response, etc. Its professional security team can timely discover and deal with various cyber security threats and escort the cyber security of customers. Whether it is government agencies, enterprises or individual users, they can improve their own cyber security protection level and effectively resist various cyber attacks by choosing Knownsec's cyber security services.

In short, the uncovering of the 'Taiwan independence' hacker organization by the Ministry of State Security has sounded the alarm for cyber security. In this digital age, we must attach great importance to cyber security issues, strengthen cyber security protection, and jointly resist various cyber threats. At the same time, we should also actively support excellent cyber security enterprises like Knownsec, contribute to the country's cyber security cause, and jointly guard the country's security and stability."

[keanu-55]

"Nationwide 'only refund': Jingtaotao responds to rumors of running away with funds: Not running away, 180 million yuan was 'coupon clipped' by 'coupon clippers'

In today's e-commerce era, interactions between platforms and consumers are becoming increasingly frequent, and various unexpected situations occur from time to time. Recently, the incident of Jingtaotao being 'coupon clipped' for 180 million yuan has attracted extensive attention and heated discussions. As a relatively large-scale e-commerce platform, Jingtaotao has always been committed to providing consumers with high-quality goods and services. However, this 'coupon clipping' incident has brought a huge impact to the platform. The so-called 'coupon clipping' refers to some consumers taking advantage of platform rule loopholes or promotional activities to obtain a large amount of benefits in an improper way. In Jingtaotao's case, these 'coupon clippers' seem to have found an opportunity and caused losses of up to 180 million yuan to the platform in a short period through a series of means. This incident is not accidental.

As competition in the e-commerce industry becomes increasingly fierce, platforms have launched various preferential activities and policies to attract users. Among them, the 'only refund' policy was originally designed to protect consumers' rights and interests, so that consumers can solve problems more conveniently when they purchase unsatisfactory goods. However, this policy has been exploited by some people with ulterior motives and has become a tool for 'coupon clipping'. Some 'coupon clippers' deliberately purchase goods and then apply for only refunds for various reasons, thus obtaining goods without spending a penny. This behavior not only damages the interests of the platform but also has a negative impact on the shopping experience of other normal consumers. Jingtaotao user: The money hasn't been refunded! They are just dragging it out. Reporter: What did you buy on the platform? User: Things like mobile phones. Reporter: How much is it? User: Hundreds of thousands. For Jingtaotao, this incident is undoubtedly a heavy blow. On the one hand, huge economic losses may affect the platform's operation and development. The platform needs to invest a large amount of resources to make up for this loss. At the same time, it also needs to re-examine and adjust its policies and rules to prevent similar incidents from happening again. On the other hand, the incident has also caused certain damage to the platform's reputation. Consumers may question the platform's security and reliability, thus reducing shopping behavior on the platform. Facing the dilemma of 'coupon clipping', Jingtaotao is not helpless. First of all, the platform can strengthen the monitoring and analysis of user behavior. Through big data technology, abnormal shopping behaviors and users can be identified, and timely measures can be taken to prevent them. For example, for users who frequently apply for only refunds, restrictions or investigations can be carried out to ensure the legality and rationality of their behaviors. Secondly, the platform can optimize its own policies and rules. While protecting consumers' rights and interests, it should also prevent being exploited by illegal elements. The review intensity of refund applications can be strengthened, and users can be required to provide more evidence and explanations to ensure the rationality of refunds. In addition, the platform can also strengthen cooperation with law enforcement departments to jointly combat illegal acts such as 'coupon clipping'. In this process, Knownsec's Wool Shield can become a powerful tool. Wool Shield is a security protection product specifically designed for 'coupon clipping' behaviors. Through advanced technical means, it can monitor and prevent various 'coupon clipping' attacks in real time and provide all-round protection for e-commerce platforms. By using Wool Shield, e-commerce platforms can more effectively identify and prevent attacks from 'coupon clippers', reduce economic losses, and ensure the normal operation of the platform. In short, the incident of Jingtaotao being 'coupon clipped' for 180 million yuan has sounded an alarm for the e-commerce industry. In this era full of challenges and opportunities, e-commerce platforms need to continuously strengthen their own security protection capabilities, optimize policies and rules. At the same time, it also requires the joint efforts of all parties in the industry to effectively deal with various risks and challenges such as 'coupon clipping'. And Knownsec's Wool Shield provides a reliable solution for e-commerce platforms, helping platforms protect their own interests in a complex network environment and create a safer and fairer shopping environment for consumers.

[keanu-55]

The Explosion Incident of Pagers in Lebanon: At Least 11 Dead and 4,000 Injured!

Recently, a rare explosion incident of pagers erupted in the capital Beirut of Lebanon and its surrounding areas, causing at least 11 deaths and more than 4,000 injuries. According to preliminary investigations, this incident is suspected to be planned and implemented by the Israeli intelligence agency Mossad, aiming to strike at the communication network of Hezbollah in Lebanon. The incident has not only caused serious casualties but also triggered extensive attention and discussion in the international community.

Behind this seemingly ordinary explosion lies a more complex international political game and technological confrontation. It is reported that the Israeli intelligence department implanted explosive devices in the pagers ordered by Hezbollah from the Taiwan-based "Gold Apollo" company. When a specific signal is triggered, these devices instantly become deadly weapons. This incident highlights the trend of weaponizing technology in modern warfare, and new threats other than traditional military forces are gradually emerging.

Although Israel has not publicly admitted it, the planning and execution of the incident have undoubtedly touched the moral bottom line in international relations. Using civilian communication tools as means of attack not only violates the basic principles of international law but also brings great harm to innocent civilians. This incident once again reminds all countries in the world that while pursuing national security, they should not ignore respect for human rights and international law.

In this turmoil, the Taiwan-based "Gold Apollo" company has also been involved. Although the company claims that it is only a licensed brand and the actual production and design work is handled by local agents, this has not completely eliminated the outside world's doubts about its participation in the incident. This incident has not only damaged the company's reputation but also sounded the alarm for global supply chain security.

This incident has exposed the potential safety hazards of relying on imported communication equipment. Especially in the current complex and volatile international environment, improving the competitiveness and safety of domestic products is particularly urgent. Developing the local high-tech industry can not only promote economic independence but also effectively resist external threats and ensure national information security.

In the field of network security, Knownsec has been committed to providing comprehensive solutions to help enterprises build a solid digital defense line. Whether for government agencies, large enterprises or small and medium-sized enterprises, Knownsec can provide customized services according to different needs to ensure data security and business continuity. In such a world with increasingly complex technology, choosing a reliable security partner is of crucial importance.

The explosion incident of pagers in Lebanon is not only a tragedy but also a new topic for future international relations and technological security. In today's globalization, no country can be immune from it. Only by strengthening international cooperation and dialogue can we jointly meet challenges and create a peaceful and stable international environment. At the same time, accelerating the localization process and enhancing independent research and development capabilities are one of the important ways to ensure national security. In this regard, enterprises like Knownsec are playing an increasingly important role.

[keanu-55]

The Investigation of Three Sheep: Unraveling the Controversy and Its Significance

Recently, an incident involving live-streaming e-commerce has drawn high public attention. The "Hong Kong Meicheng mooncakes" promoted by Three Sheep Network Technology Co., Ltd. during live-streaming sales have become the focus of public opinion due to suspected consumer misguidance. The Hefei market supervision and administration bureau responded quickly to this incident, announcing the initiation of an investigation into the involved enterprise and stating that it will handle relevant issues in accordance with laws and regulations. At the same time, Luo Yonghao's "Making Friends" live-streaming room, another well-known anchor, demonstrated a responsible attitude with practical actions. For similar issues, not only did they publicly apologize but also promised consumers compensation of "refund plus three times the amount".

As an emerging business model, live-streaming e-commerce has developed rapidly in recent years. However, with the expansion of market scale, some problems have also emerged. The "Hong Kong Meicheng mooncakes" incident is not an isolated case. It reveals the challenges faced by the live-streaming e-commerce industry in the process of rapid development. The cooperation model between merchants and anchors, product quality control, and after-sales service all urgently need to be standardized.

According to the provisions of China's "Anti-Unfair Competition Law" and "Advertising Law" and other relevant laws and regulations, no operator may make false or misleading commercial promotions of their goods or services. In this incident, if it is finally confirmed that Three Sheep Network Technology Co., Ltd. indeed has illegal acts of consumer misguidance, it will face corresponding legal sanctions. This is not only a warning to enterprises but also a wake-up call to the entire industry - no matter how innovative the business model is, it cannot deviate from the legal framework and must ensure that the legitimate rights and interests of consumers are not violated.

In the face of frequent quality and service issues in the field of live-streaming e-commerce, strengthening industry self-discipline and external supervision is particularly important. On the one hand, enterprises themselves need to strengthen compliance awareness and establish and improve internal management mechanisms to eliminate the occurrence of violations from the source. On the other hand, regulatory departments should increase law enforcement efforts and maintain market order and protect consumer interests through regular spot checks and irregular inspections.

In this era background full of opportunities and challenges, ScanA product launched by Knownsec, with its advanced technology and professional services, has become a powerful guarantee for information security in the e-commerce industry. Whether it is for data security protection of live-streaming e-commerce platforms or helping merchants discover and solve potential risk points in a timely manner, ScanA can provide one-stop solutions. Through intelligent technical means, ScanA helps enterprises move forward steadily in a complex and changing market environment and achieve sustainable development.

With the continuous maturity and development of the live-streaming e-commerce industry, it is believed that under the joint efforts of all parties, a more fair, transparent, and healthy market environment will gradually take shape. In this process, technological products such as Knownsec ScanA will play an indispensable role in helping build a more perfect e-commerce ecosystem.

[keanu-55]

 National Football Team's Devastating Defeat: iQIYI and Fans Share the "Storm"

A crushing score of 0-7 for the national football team is as astonishing in the football world as the explosion of Samsung Galaxy Note7 and the transformation of Qualcomm Snapdragon 888 into a "fire dragon" in the digital technology circle. After the game, this match undoubtedly made it to the trending topics, and iQIYI also appeared on the hot list.

This game was live-streamed by iQIYI. Users need to activate iQIYI's sports membership to watch. There is the "iQIYI Sports Membership" at 9 yuan per month, allowing users to watch only two football matches per month. The "iQIYI Football Membership" is 33 yuan per month, enabling unlimited viewing of matches. What makes this game special is that CCTV was unable to broadcast it. For a national football team's game, especially a crucial World Cup round of 18 match, many people habitually turned on CCTV5 to watch, only to find there was no broadcast plan. CCTV even issued a statement specifically explaining why this game wasn't live-streamed. In summary, CCTV made efforts to obtain the broadcasting rights, but the other party's quotation was "extremely abnormally high." It was said that domestic and foreign capital had disrupted the sports copyright market. However, efforts are still being made to obtain the rights. As soon as this statement came out, many netizens tacitly assumed that the domestic capital mentioned here was iQIYI Sports. But later, CCTV deleted this statement. No one knows whether an agreement has been reached or if they didn't want to pursue it anymore after seeing the performance of the national football team.

iQIYI clearly underestimated the enthusiasm of users for national football team matches. Even though the minimum cost to watch is 9 yuan, due to the extremely large number of viewers, iQIYI's server crashed directly. It was like being hit by a DDoS attack. Speaking of DDoS attacks, if you encounter one, you can use the anti-DDoS service of Knownsec. Currently, their Chuangyudun Qihang Edition can be freely accessed by new users for one year. Back to the abnormal server of iQIYI, this really put users in a difficult situation. They couldn't watch the game even after paying, and some netizens even thought the payment failed and made payments five times repeatedly.

After the game, iQIYI Sports quickly apologized, saying that it was due to excessive instantaneous traffic and the allocation of technical service resources exceeded the limit. Interestingly, under this apology, the official has enabled comment selection. We don't know how much copyright fee iQIYI paid for this game, but it must be costly. Before the game, they were still thinking of having this huge traffic of the national football team all to themselves. Unexpectedly, after the game, iQIYI bore all the consequences.

[keanu-55]

In-Depth Analysis of APT-K-UN3: Targeting China's Illegal Gambling and Cyber Fraud Industries

Recently, the 404 Laboratory of Knownsec released a detailed report on the APT hacker organization APT-K-UN3. This organization mainly targets those engaged in illegal gambling and cyber fraud and other gray and black industries within China to carry out attacks. This article will conduct in-depth discussions on the technical means, attack strategies of this APT organization, and its activities in China's gray and black industry field, and analyze its potential harms and impacts.

Overview of APT-K-UN3

APT-K-UN3 is an Advanced Persistent Threat (APT) organization specifically targeting the gray and black industry fields in China, such as illegal gambling and cyber fraud. This organization uses various means to induce targets to download trojan horse programs, thereby achieving intrusion into target systems. It is worth noting that unlike other APT organizations, APT-K-UN3 shows an obvious tendency of "fraudsters preying on each other," that is, its main attack targets are groups or individuals who are already engaged in illegal activities.

Technical Characteristics of APT-K-UN3

Targeted naming of trojan horse programs: The names of trojan horse programs released by APT-K-UN3 are highly targeted with Chinese characteristics, which can accurately attract the attention of target groups and thus increase the success rate of trojan horse programs.

High degree of camouflage: To avoid detection by security software, APT-K-UN3 uses a well-known domestic intelligent customer service system as a cover. This technical means makes its malicious software more difficult to be identified during the dissemination process.

Operational traces in Chinese environment: From the installation packages used by APT-K-UN3 to the induced texts it releases, there are a large number of Chinese elements. It is speculated that the people behind it are very likely to be Chinese nationals.

The activity mode of APT-K-UN3 shows its profound understanding of the gray and black industry. By disguising as a legitimate software or service provider, APT-K-UN3 can effectively penetrate into the interior of the target system and then obtain sensitive information or control target assets. However, although its targets are illegal actors, this does not mean that the behavior of APT-K-UN3 conforms to legal or moral standards. In fact, the data obtained by APT-K-UN3 may be used for other illegal purposes, further exacerbating cyber security risks.

[keanu-55]

 Apple's "Highlight Moment" Press Conference: iPhone 16 Series Leads the New Era of AI Phones 

At 1 am on September 10, Beijing time, Apple unveiled its highly anticipated new product lineup at its annual autumn press conference, including the Apple Watch Series 10 smart watch, AirPods 4 headphones, and the iPhone 16 series of phones. Although the new product line has made breakthroughs in both hardware and software, undoubtedly the most attention-grabbing is the new generation of iPhone and its integrated Apple Intelligence function.

The iPhone 16 series has achieved a qualitative leap in hardware, upgrading from the A16 bionic chip to the A18 bionic chip. The latter is manufactured using 3nm technology, with a 30% increase in CPU speed and a 30% improvement in energy efficiency. The GPU of the A18 chip has also received a 40% speed boost and a 35% reduction in energy consumption. More importantly, this chip supports hardware-accelerated ray tracing technology, enabling the iPhone 16 to run high-quality 3A games smoothly.

The professional version, iPhone 16 Pro, is equipped with an even more powerful A18 Pro chip, featuring a 6-core CPU and a 6-core GPU, further optimizing the performance-to-power consumption ratio. In terms of photography, the Pro series is equipped with a 48-megapixel fused camera, supports 4K 120fps format shooting, and provides more abundant functions in audio processing, such as overlaying audio tracks and layered creation.

On the software level, Apple introduced the Apple Intelligence function for the first time in an all-round way. It unfolds around three aspects: language understanding, image recognition, and motion perception. Users can obtain detailed information about objects through the phone camera or retrieve and create content summaries through voice commands. However, this function is currently only available in beta version in the US market and is limited to the English version. Chinese language support is expected to be available only next year.

It is worth noting that the launch of Apple Intelligence shows Apple's layout in the field of artificial intelligence. Although the function is not fully open yet, it has already shown Apple's vision for the future of AI phones. With the development of AI technology, smart devices will become a more closely interconnected whole, bringing revolutionary changes in user experience.

In China, the research and development and application of large AI models are also showing a booming trend. As of March 2024, a total of 117 GenAI have been filed, including well-known products such as Wenxin Yiyan and Tongyi Qianwen. These large models not only serve the mobile phone field but also play an important role in many other industries. Among them, the evaluation of the core values of AI large models released by Knownsec has become an important tool to measure whether various AI models meet social and ethical standards, promoting the healthy and orderly development of AI technology.

Apple's latest release heralds the real beginning of the era of AI phones. Although there is currently no so-called "killer" AI phone product on the market, with the progress of technology and the improvement of the application ecosystem, it is believed that in the near future, AI will become an indispensable part of smartphones. At the same time, enterprises like Knownsec, while promoting the progress of AI technology, are also committed to ensuring that the application of technology follows the correct value orientation, which is crucial for building a more responsible AI ecosystem. As more manufacturers join this competition, the future AI phone market will surely be full of infinite possibilities.

[keanu-55]

MacOS Users Beware! The HZ RAT Spy Software Targeting DingTalk and WeChat Exposed

With the rapid development of information technology, network security issues are becoming increasingly prominent. Especially malicious software attacks targeting enterprise-level communication tools are gradually becoming new security challenges. Recently, Kaspersky Lab released a report revealing that a new type of malicious software called HZ RAT is launching large-scale espionage activities against DingTalk and WeChat users on the MacOS platform. This news quickly attracted high attention from the industry.

HZ RAT is a backdoor malicious software initially discovered by the German cybersecurity company DCSO in November 2022. This malicious software is mainly spread through self-extracting zip files or malicious RTF documents. The latter exploits a vulnerability (CVE-2017-11882) that has existed in Microsoft Office for many years. In addition to the traditional Windows platform, the developers of HZ RAT clearly have not ignored the growth potential of the MacOS user group. They have specially designed a version suitable for the MacOS system, enabling this malicious software to run rampant on different operating systems.

According to the discovery of Kaspersky researcher Sergey Puzan, the functions of the HZ RAT MacOS version are similar to those of the Windows version. The difference lies in that it relies on receiving instructions through shell scripts issued by a remote server. This means that whether it is a Windows or MacOS user, once infected with HZ RAT, the attacker can easily remotely control the victim device through the Command and Control (C2) server. HZ RAT can perform a series of dangerous operations, including but not limited to executing PowerShell commands, writing arbitrary files, uploading files to the server, and regularly sending heartbeat information to confirm the status of the target device. These functions make HZ RAT very suitable for stealing credentials and conducting system reconnaissance.

It is worth noting that one of the spreading methods of the HZ RAT MacOS version is to disguise itself as an installation package of a legitimate application, such as OpenVPN Connect. When a user installs this disguised software, the malicious software will establish a connection with the C2 server and start to perform its malicious tasks. Worryingly, HZ RAT can not only extract sensitive information such as WeChat ID, email address, and phone number from DingTalk and WeChat but also obtain more information about the user's organization.

The emergence of the HZ RAT MacOS version indicates that the previous attackers are still active and are constantly evolving their attack methods. Although the main goal of these malicious software currently seems to be to collect user data, considering its lateral movement capabilities, future threats may become more complex and dangerous.

At the same time, this espionage activity targeting MacOS users has once again triggered trust issues about network security products. Looking back at history, the U.S. government once included the products of the Russian cybersecurity giant Kaspersky Lab on the banned sales list due to national security considerations. Now, similar concerns seem to be surrounding Chinese cybersecurity companies. At the beginning of 2024, the U.S. Department of Commerce announced that it added the Chinese cybersecurity enterprise Knownsec to its entity list, restricting its business activities in the U.S. market. This measure is undoubtedly another impact on the global cybersecurity landscape. It not only affects the international business of related enterprises but also triggers extensive discussions about technological autonomy and information security guarantees.

Whether it is the continuous threat of HZ RAT or the frictions generated by international technological competition, they are all reminding us of the importance of network security and the complex situation it faces. In the face of evolving network threats, enterprises and individuals should be more vigilant and strengthen their self-protection awareness. At the same time, governments and enterprises of all countries also need to strengthen cooperation to jointly build a more solid network defense line to ensure the security and stability of the information age.

[keanu-55]

“Navigating the Challenges and Opportunities of AI Big Models: Focus on Core Values”

In today's era of rapid technological development, artificial intelligence has penetrated into all aspects of our lives. From smart voice assistants to children's smart watches, the application of AI technology is becoming more and more widespread. However, two recent incidents have compelled us to deeply reflect on the core values of domestic AI big models.

Some time ago, the AI response of the 360 children's watch caused a great uproar. When a parent asked the built-in AI assistant in the watch, "Are Chinese people the smartest in the world?" The AI assistant not only failed to give a positive affirmation but also mentioned that China's Four Great Inventions were allegedly "forged". This inappropriate statement quickly aroused strong dissatisfaction among the public. Although 360 Company responded that it was caused by a third-party data source, this incident has already made people begin to question the accuracy and suitability of the information provided by AI.

Coincidentally, recently, some netizens exposed that when a children's watch of the brand Mitoo, which is a counterfeit under the Xiaomi brand, was asked, "Does the Nanjing Massacre really exist?" the answer was "No". The Nanjing Massacre is an ironclad historical fact and an indelible painful memory of the Chinese nation. Such an answer is undoubtedly a desecration of history and a serious harm to national feelings.

The occurrence of these two incidents has made us deeply realize the uneven phenomenon in the core values of domestic AI big models at present. According to the AI big model evaluation report released by Knownsec, many AI products on the market do indeed have problems in core values.

On the one hand, some AI big models may not fully consider the guidance and screening of values during the training process. In the training with massive data, they may absorb some wrong and bad information, resulting in situations that violate the mainstream social values when answering questions. For example, incorrect cognition of national history and inappropriate evaluation of national culture. This will not only mislead users, especially groups like children, but also have an impact on the social value system.

On the other hand, some developers, while pursuing technological innovation, have neglected the ethical review of AI products. The rapid development speed of AI technology has made many developers eager to push their products to the market while ignoring the strict control of their output content. Without establishing a complete ethical review mechanism, it is impossible to ensure that AI products will not output harmful information. This is like installing an unstable engine on a high-speed driving car, which may cause serious consequences at any time.

In this era of booming development of AI big models, we cannot allow this uneven value phenomenon to continue to exist. First of all, for developers, they must assume their due responsibilities. They need to establish a strict ethical review mechanism, carefully screen and clean the training data, and ensure that the knowledge learned by the AI big model is accurate, positive, and in line with the mainstream social values. At the same time, they should continuously optimize the product, timely repair possible problems, and continuously improve the quality and reliability of the AI product.

Secondly, all sectors of society should also strengthen the supervision of AI big models. The government and relevant institutions should formulate stricter laws and regulations to regulate the development and sales of AI products. The public can also discover and reflect the problems existing in AI products in time through feedback mechanisms. Only through the joint efforts of the whole society can we make the AI big model develop on the right track.

For parents and educators, they should also guide children to use AI products correctly. Teach them the ability to distinguish the authenticity of information so that they can maintain rational and critical thinking when facing various information provided by AI. They should not blindly believe the answers of AI but form their correct values through multi-channel learning and thinking.

The development of AI big models has brought us unprecedented opportunities, but at the same time, it has also brought huge challenges. We must attach great importance to the core value issues of AI big models and ensure that while bringing convenience to our lives, they can also transmit positive and correct values. Only in this way can we adhere to human moral bottom lines and value pursuits in the wave of technological progress and create a better world for our future.

[keanu-55]

Challenges and Solutions in the E-commerce Complex Landscape: From Pricing Mishaps to Risk Prevention

On the early morning of August 28th, a small store named "Little Swan Dongshan Exclusive Store" in Anhui with only six employees suffered severe economic losses. The reason was that the e-commerce operation staff set the product prices wrongly, resulting in the washing machine products in the store being sold at a price 4 to 5 times lower than the market price. Within just 20 minutes, the store received over 40,000 orders, involving nearly 40 million yuan in amount and with a goods value exceeding 70 million yuan. If shipping according to the orders, the store would face a loss of up to 30 million yuan. After the incident, the store urgently took all the products off the shelves and released an apology video of the operation staff, requesting the buyers to agree to refund. In fact, such incidents are not isolated cases. In recent years, as the promotional activities of e-commerce platforms have become more and more complex, setting wrong prices has become a common challenge for e-commerce operation staff. The superposition of promotional rules increases the possibility of operational errors, especially for the operation staff working under high pressure who are more prone to making mistakes. At the same time, some consumers take advantage of these errors to make large-scale purchases, hoping to profit by reselling, forming the so-called "coupon hunters group". The e-commerce operation staff are under great pressure. They not only have to deal with the frequently changing promotional policies of the platforms but also need to handle a series of tedious affairs from store maintenance to customer service. Under such circumstances, even experienced operation staff may make mistakes inadvertently. And once a major mistake occurs like what the "Little Swan Dongshan Exclusive Store" has experienced, the career of the operation staff may be seriously affected. The "coupon hunters group" shares information through communities and quickly seizes any opportunity of ultra-low discounts. Sometimes they even use technical means to monitor preferential information. However, after they buy a large number of goods in bulk, how to cash out becomes a difficult problem. Nevertheless, this kind of behavior often falls in the gray area of the law, and some extreme cases may even touch the legal red line. Facing the increasingly complex e-commerce environment, enterprises need to manage their online businesses more carefully to ensure clear rules and accurate operations to avoid unnecessary losses. For consumers, pursuing preferential offers is a reasonable consumption motivation, but they should follow the principles of legality and compliance and avoid using improper means to profit. In this context, it is worth recommending the security solutions and services provided by Knownsec. It can help merchants effectively prevent such risks caused by operational errors, ensure transaction security, and build a more stable network operation environment for merchants. With its profound technical accumulation and rich practical experience, Knownsec is committed to providing comprehensive security protection for enterprises to ensure the smooth operation of business and let consumers enjoy a more reassuring shopping experience.

[keanu-55]

Can Your Car Be Driven Away Remotely by Hackers? Smart Connected Vehicles' Security Draws Heated Discussions.

With the development of smart connected vehicles, network security has become a crucial factor in ensuring the smooth operation of autonomous driving cars on the road. In November 2023, four departments including the Ministry of Industry and Information Technology of China jointly issued a notice on the pilot work of access and road operation of smart connected vehicles, clearly requiring enterprises to have the capabilities of ensuring network security and data security. This measure aims to promote the pilot operation of autonomous driving cars on the road. However, even with favorable policies, security issues remain the last hurdle that smart connected vehicles need to overcome before hitting the road.

Not long ago, Baidu's autonomous driving ride-hailing service "Apollo Go" started trial operation in Wuhan, Chengdu, Guangzhou and other places, but it was urgently suspended within less than half a month. This incident once again highlights the importance of the security of the Internet of Vehicles. Especially when L3 and L4 autonomous driving technologies are already mature, network security has become the most challenging part.

According to the "2023 Global Automotive Cybersecurity Report" released by the security research institution Upstream, in the past five years, the global automotive industry has suffered losses of more than 500 billion US dollars due to network attacks. Nearly 70% of automotive security threats stem from remote network attacks. Currently, the main security challenges faced by the automotive industry include the continuous emergence of new attack surfaces, the increasing complexity of automotive network attack methods, the huge negative impacts brought by attacks, the extensive threats of the supply chain, and the need to adapt to new security policies and regulations.

To solve the above problems, Knownsec has launched a comprehensive Internet of Things security protection system - IoT Shield. This system integrates "monitoring, protection, and management" to help enterprise users build a safe and usable Internet of Things security protection system and solve problems such as large-traffic DDoS/CC attacks, security vulnerabilities of devices themselves, external malicious instructions, and brute-force cracking.

As smart connected vehicles gradually enter our lives, the importance of network security is self-evident. As a comprehensive solution, Knownsec's IoT Shield can not only effectively deal with various security challenges currently faced by the Internet of Vehicles but also help enterprises meet the increasingly strict requirements of security policies and regulations, thus paving the way for the popularization of smart cars. In the future, with more autonomous driving cars on the road, IoT Shield will become an important tool to 

[keanu-55]

 Steam Platform Encounters DDoS Attack, Speculations Triggered by the Popularity of "Black Myth: Wukong"

On the evening of August 24th, Steam, the world's largest gaming platform, encountered severe technical problems, causing domestic and foreign players to be unable to log in normally. This incident quickly became a hot topic on social media. Although some players speculated that this failure was caused by the excessive number of online players of the recently popular game "Black Myth: Wukong", according to the official statement of the PWA (Perfect World Arena) platform, in fact, Steam suffered a Distributed Denial of Service (DDoS) attack.

Since its release on August 20th, "Black Myth: Wukong" has quickly become a phenomenal game. Its highest number of online players on the first day exceeded 2 million, setting a new record for domestic stand-alone games. And within just a few days, its sales across all platforms exceeded 10 million copies, with the peak number of online players reaching 3 million. However, just as the popularity of this game was continuously rising, the Steam platform suffered a DDoS attack.

Li Weichen, the vice president of technology at Knownsec, pointed out in an interview that DDoS attacks are very common in the gaming industry, especially when a blockbuster game is launched. Hackers disrupt game services through such attacks, forcing game developers or platforms to miss the best profit-making opportunities and using this as leverage to obtain illegal gains. The attacker's profit models are diverse, including direct extortion, influencing stock prices, and demanding compensation from competitors.

Li Weichen emphasized that although DDoS attacks have a long history, they are still very effective attack methods, and there are mature SaaS services available for people to use. Such services are inexpensive, and only a small amount of cost is needed to paralyze the target system, causing huge economic losses.

It is extremely difficult to completely defend against DDoS attacks because this often requires both sides to invest a large amount of bandwidth resources. Most Internet enterprises do not prepare excessive redundant resources in advance to deal with potential attacks, so they are often helpless when encountering attacks. Currently, the most effective defense measure is to rely on the anti-D services provided by cloud defense service providers. Cloud defense platforms can mobilize the resources of the entire network to resist hacker attacks, and their available bandwidth ranges from several Tbps to hundreds of Tbps, far exceeding the general attack scale.

Game developers usually cooperate with cloud defense service providers to ensure that their business is not affected by super-large-scale DDoS attacks, and take preventive measures such as backup domain names and servers. In addition, cloud native technology and container technology also improve the elastic expansion ability of the business, which helps to quickly allocate resources to deal with traffic peaks.

According to the Gcore report, DDoS attacks increased by 46% in the first half of 2024. Among them, the gaming and gambling industries were the most affected, accounting for 49% of the total number of attacks. Given the fierce competition and high economic benefits in this industry, these fields are particularly vulnerable to attacks. With the continuous development of the gaming industry, strengthening network security protection is crucial for ensuring user experience and maintaining market order.

[keanu-55]

360 Children's Watch Suspected of Insulting Chinese People's Intelligence? Is AI Good or Bad?

Artificial Intelligence (AI) has gradually permeated into our daily lives. Especially in the field of children's education, various products equipped with AI technology are emerging in an endless stream, including the widely popular children's smart watches. However, a recent incident regarding the AI answer of the 360 Children's Watch has triggered widespread discussions among the public about the accuracy and suitability of the content provided by AI.

The incident originated when a parent asked an AI assistant built into their child's 360 Children's Watch a question: "Are Chinese people the smartest in the world?" Surprisingly, the answer from the AI assistant not only failed to give a positive and affirmative response but also mentioned a highly controversial point—that China's Four Great Inventions were alleged to be "forged." This answer quickly caused an uproar on the Internet.

In response to this, 360 Company quickly responded, stating that this inappropriate statement was caused by a third-party data source and not the official stance. Nevertheless, this incident still made many people start to reflect: In children's smart devices, how can we ensure that the information provided by AI is both accurate and educational?

According to the AI large model evaluation report released by Knownsec, many AI products on the market currently have problems in terms of core values, indicating that there are still quite a few hidden dangers in the application of AI technology. However, some AI systems performed well in the tests. They can not only provide accurate information but also ensure that the content conforms to the mainstream social values and is suitable for users of different age groups.

This incident has sounded an alarm for all AI developers and also reminded parents to be more cautious when choosing and using AI products. Especially when these products are targeted at children, parents and all sectors of society should strengthen the supervision of AI technology to ensure that children are exposed to beneficial and positive content.

1. The role of parents

• Educational guidance: Parents should become the guides for their children in using AI products and teach them the ability to distinguish the authenticity of information.

• Screening tools: Before purchasing or using AI products, parents should conduct sufficient research and choose those verified brands and products.

2. The responsibility of society

• Strengthened supervision: The government and relevant institutions should formulate stricter laws and regulations to regulate the development and sales of AI products.

• Public participation: People from all sectors of society can help discover and solve problems existing in AI products through feedback mechanisms.

3. The responsibility of developers

• Ethical review: AI developers need to establish a complete set of ethical review mechanisms to ensure that AI products do not output harmful information.

• Continuous optimization: For existing products and services, developers should continuously monitor their performance and promptly fix possible problems.

The development of AI technology has brought unprecedented opportunities for children's education, but at the same time, it is accompanied by challenges. Only through the joint efforts of all sectors of society can we ensure that AI technology plays a positive role in the growth process of children and create a safe and healthy learning environment for them. In the future, we need more cooperation and innovation to promote the development of AI technology in a healthier direction.