Hacker News Reader

Comment by vintermann | Hacker News Reader

Pretty crazy. The security team denies there ever was a threat of a strike? And generally you do announce strike threats, it's kind of the whole point with strikes that your employer knows of the possibility.

Did Anthropic get "warned" about this strike by its own AI models?

View on HN

It's possible that they just took the state of contract negotiations as an implicit threat of a strike without waiting for the union to explicitly announce the threat.

I'm assuming something along the lines of:

"based upon the intercepted emails and language used in discussions with the union, we predict with 63% accuracy that they will strike within ~3 months"

just because AI is really good at being wrong doesn't mean that people aren't themselves capable of being wrong manually. Do I think someone may have consulted AI? Almost certainly, in fact I'd guess that there's an internal policy document somewhere saying that they have to consult AI before making choices like this. Do I think that good old fashioned human incompetence and baseless speculation are at the heart of this? also yes.

Well isn't that a damn terrible idea? Who convinced them that was a good idea? "We have a heard faint rumors of a whiff that you may one day threaten a strike, so we're gonna try insane overreaction as a tactic"

a week of wfh from a company that's equipped to support remote work anyway and only wants in-office 25% of the time hardly qualifies as "insane overreaction". I'd says it's probably more along the lines of "a single email".

Reply on news.ycombinator.com