Following a link in Martin Fowler’s Fragments, I stumbled upon Genie Tarpit by Kent Beck – a perfect summary of my experiences with AI coding (code reviews are OK, new code less so). He also provided a good reason for that behavior:

The “plausible deniability” task orientation of the genie leaves it claiming success even though the code doesn’t work at all.

And the proposed solution?

You probably saw this one coming—nobody knows.

Yeah, that about sums it up.