I think the NA28 got I Jn 5:18 wrong, and here's why.
Don’t know why I haven’t spotted this before, but the NA28 text has changed in I Jn 5:18 from Οἴδαμεν ὅτι πᾶς ὁ γεγεννημένος ἐκ τοῦ θεοῦ οὐχ ἁμαρτάνει, ἀλλ᾿ ὁ γεννηθεὶς ἐκ τοῦ θεοῦ τηρεῖ αὐτὸν καὶ ὁ πονηρὸς οὐχ ἅπτεται αὐτοῦ. (NA27)
to
Οἴδαμεν ὅτι πᾶς ὁ γεγεννημένος ἐκ τοῦ θεοῦ οὐχ ἁμαρτάνει, ἀλλʼ ὁ γεννηθεὶς ἐκ τοῦ θεοῦ τηρεῖ ἐαυτόν καὶ ὁ πονηρὸς οὐχ ἅπτεται αὐτοῦ. (NA28)
The big difference is in the reflexive pronoun replacing the personal pronoun in the phrase, “the one who was born of God keeps him/himself”. αὐτὸν changes to ἐαυτόν. This brings the NA text into agreement with Scrivener’s Textus Receptus (and other Byzantinish texts), except for the punctuation.
The interpretation of this verse involves considering who ‘the one who has been born of God’ (ὁ γεγεννημένος) and ‘the one who was born of God’ (ὁ γεννηθεὶς) are. Are they both the born again Christian, or is the ὁ γεννηθεὶς Christ? In John’s writing he often uses the participle of γενναω to refer to the believer, but always in the perfect passive tense (ὁ γεγεννημένος), not aorist (ὁ γεννηθεὶς). Which argues against the reflexive pronoun. So, as far as intrinsic probabilities, NA27 looks more obvious.
Sadly, for such a big change (and one of few made to the entire text), the NA28 offers precious testimonial evidence in the apparatus. Having only “αυτον A* B 1852 latt”. Not even a positive argument for εαυτον. That’s why you keep your old editions. NA27 has for the reflexive pronoun: 01 A(c) P Ψ 33. 1739 Majority; Origen The problem with the witnesses for αυτον, however, is that they can’t all be counted as witnesses for ὁ γεννηθεὶς … αὐτὸν. Both the 1852 as well as latt are testifying to η γεννησις … αυτον, “the birth which is from God keeps him”. This is a completely different reading. The NA28 apparatus notes this difficulty: when citing the witnesses for η γεννησις it recommends the reader compare the witnesses for αυτον.
The witnesses for η γεννησις are too weak to suppose it original. So, many good witnesses for the personal pronoun must go, because they aren’t really witnesses for the personal pronoun, but a different reading. The remaining witnesses seem impressively to favor the reflexive pronoun when a simple head count is done. Even beyond that, the spread of location and age make it a good candidate for original, considering external evidence alone.
But external evidence, though much, is not all. Given that the two participles differ in tense, and the Johannine style in referencing believers as ὁ γεγεννημένος, perfect passive participle, and never with the aorist tense, leaves too much question to simply capitulate to the majority of manuscripts. Further, the context of the verse is the preceding verses 15-17, wherein we are told to pray for those brothers who have sinned (but not sinning the sin unto death) and God will give them life. To further explain that deliverance comes from Christ’s keeping him that has been born of God, in answer to your prayers, jibes better, I think, than changing the subject to suddenly explain that believers don’t sin anyhow because they keep themselves from the evil one.
So the personal pronoun does have Greek manuscript support (and Alexandrinus teamed up with Vaticanus together is nothing to sneeze at), it has Johannine style backing it, it has the natural force of the grammar behind it, and it has the logical flow of the text. I believe ὁ γεννηθεὶς ἐκ τοῦ θεοῦ τηρεῖ αὐτὸν remains the best text and most likely original, contrary to the NA28s judgment.