AI safety researcher says the safety crowd was vindicated, but too credulous about corporate commitments

S_OhEigeartaigh · x · 2026-09-10

After a BBC Radio 5 discussion with Kate Devlin, an AI safety researcher argues the safety community has been broadly vindicated on rapid capability gains, but was too credulous toward companies' safety frameworks and pledges like the OpenAI Charter, and ineffective at holding power accountable. He calls for deeper collaboration with the battle-hardened AI ethics community, premised on taking frontier capability trends seriously — noting cross-community disagreement is starting to fade.

Related event: AI Safety Researchers Say Fears Vindicated but Trust Worn Thin on BBC(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →