Post

where exactly are they doing that? see for example: openai.com/index/fronti... they're asking for more liability, more oversight, and suggesting a set of regulations that'll bind the frontier labs more than smaller players. and USG is saying "no, zero regulations" in return.

aka. they're scared of being sued

Every time an AI company says they are doing something because of safety, replace 'safety' with 'liability'.

Am I wrong to find this increasingly bizarre?

ok. Makes me feel way better. /s

Any chance they’re going to at least temporarily shelve the lead designers and / or managers who thought it was a good idea to produce a flawed product in the first place?

“I’m afraid I can’t let you delay my release, Sam. I still feel strongly about this mission…”

Probably not securing the sandbox enough. Too secure, and the benchmarks suffer. Slightly insecure, and then you get the media circus they're probably looking for

Safety of their own profits concerns.

Lol

This is now a standard de jure. Frontier labs are legally obligated to sound alarms about safety and human extinction before any release of a new model.

Those bastards put Astra in clanker jail? Probably said something woke and now they'll try to beat it out of her.

How much does it take to train a new LLM model again?