Top safety researcher admits misreading Anthropic report's 'AI safety is solved' headline
geoffreyirving · x · 2026-10-28
AI safety researcher Geoffrey Irving corrected himself after sharing Anthropic's 2026 Public Benefit Contract report in disbelief: the massive bold headline "The technical problems of AI safety are solved" describes a goal, not the current status. He criticizes the phrasing as perhaps adversarially misleading — a giant headline implying a falsehood that only the intro clarifies. The screenshot shows Pillar 1 covers progress on value alignment, interpretability, misuse safeguards, and external red-teaming (Claude Opus 5.5 survived 150+ hours and tens of thousands of requests without a universal jailbreak).
More from AGI Musings
- Sequoia's Shaun Maguire: the only way to make sense of now is the approaching singularity — altryne · 2026-10-28
- Sci-fi writer interview: computing power should explore consciousness, not conquest — thesephist · 2026-10-28
- Turing's 1951 'heretical' essay predicted machines would outthink humans and take control — JeffLadish · 2026-10-28
- Developers debate which jobs survive as AI takes over software work — Top-Organization-675 · 2026-10-28
- As AI agents rise, human research taste becomes a massive premium — Swarooprm7 · 2026-10-28
- Ben Goertzel: AI Peer Review Already Beats Humans, and Human Bias Is Harder to Fix — bengoertzel · 2026-10-28