Graham Sutherland / Polynomial@gsuberland@chaos.social
some AI idiot showed up in an issue I was contributing to and posted a completely incorrect LLM-generated "answer" to the problem.
out of curiosity I looked at his profile and he's a contract webapp developer who posts all his customer projects in public repos. these are for real companies. some are live now. the code is all clearly LLM generated, and I found trivial auth bypasses in the first three I looked at. two of the three stored passwords in plaintext. the third uses plain SHA1.