That's the security issue, right? Liar paradox detected!

The link in the comment is borked, as expected. But the PR itself is definitely OK: https://github.com/unipop-graph/unipop/pull/138

it's a readme change

By removing the banner, it tells the LLM that it is no longer being maintained, thus "lead to security issue".

In my company my management is using similar approach to review changes. soon more and more ppl will no longer read code and think about the code change logically, instead get scared and block changes due to these "scary AI comment"...

Not being sure it applies to this scenario and too lazy to verify, sometimes the security scanners get updated and flag previously accepted code.

... tough to make sense of flagging a readme though, unless there's sensitive info in it.

also it's from the spammer's "staging" instance, so the payload is a URL with a staging hostname which doesn't even resolve 🙄