Post #4206580
2026-07-29 16:55 UTC
1,100 employees of OpenAI, Anthropic, Google, Meta and other companies asking the U.S. government to slow down AI development because there is “a real risk” that AI progresses faster than people can “understand or control.”
My $0.02 is this wouldn't be needed if people could be trusted not to use unreliable LLMs in critical situations or at least appropriately sandbox them. But it seems even OpenAI can't figure out how to do that.
https://www.theverge.com/ai-artificial-intelligence/972161/ai-leaders-us-government-openai-anthropic-google-meta
Replies (4)
-
@famousringo@mstdn.ca 2026-07-29 18:00
@carnage4life@mas.to local Ferrari owners ask government to do something about all these people speeding around in sports cars.
-
@nf3xn@mastodon.social 2026-07-29 16:58
@carnage4life@mas.to The risk from AI is not because they could be intelligent it is precisely because they are not intelligent. Runaway robots do dumb shit.
-
@noah@23.social 2026-07-29 17:01
@carnage4life@mas.to "appropriately sandbox them" the problem is as we have seen, this isn't even enough
-
@flq@freiburg.social 2026-07-29 18:00
@carnage4life@mas.to so half a century+ of IT systems development has not taught people that there is no foolproof safety net, ever?