Anthropic's Clark moves recursive AI self-improvement forecast to 2027
Is this a scandal?
Not yet — an early signal. Noise 35/100, cooling down, across 3 sources.
Safety labs will likely publish updated RSI risk assessments within six months because Clark’s specific benchmark claims invite independent verification and competitive response.
Noise 35/100 — louder than 96% of tracked AI controversies.
Why it matters
Accelerating RSI forecasts compress the window for safety alignment and governance before models potentially exceed human research capabilities.
Key points
- Jack Clark moved his full recursive self-improvement forecast from 2028 to 2027 based on new capability data.
- The TasteVal benchmark allegedly shows AI research intuition doubling every three months since December 2025.
- Clark claims Claude Opus 5.5 outperforms human experts in research tasks at 1/30th the cost per trial.
- Full RSI implies AI systems could autonomously enhance capabilities beyond direct human control or comprehension.
- The accelerated timeline reduces the available window for safety alignment and regulatory preparation.
The story
Anthropic co-founder Jack Clark has revised his forecast for full recursive self-improvement (RSI) in artificial intelligence from 2028 to 2027. Clark cited the new TasteVal benchmark as evidence that AI research intuition is doubling approximately every three months since December 2025. He stated that Claude Opus 5.5 now surpasses human experts in research tasks while operating at roughly one-thirtieth of the cost per trial. This acceleration suggests AI systems may soon autonomously enhance their own capabilities without human intervention. The updated timeline implies a significantly shortened period for implementing safety measures before potential capability overhang. Industry observers note this marks one of the most specific near-term RSI predictions from a leading safety-focused lab executive.
Who's involved
Co-founder, Anthropic
Claims empirical benchmark data justifies accelerating RSI timeline to 2027 due to rapidly compounding AI research capabilities.
Amplified Clark’s forecast to Japanese-speaking audiences while highlighting the implication that self-improving AI evolution may become unstoppable.
How the conversation shifted
Polarity (0–100) from the noise pipeline, sampled over time.
Noise Level
The timeline
Clark announces revised 2027 RSI forecast
Anthropic co-founder publicly advances recursive self-improvement timeline citing TasteVal benchmark and Claude Opus 5.5 performance.
AI research intuition baseline established
Clark identifies December 2025 as the starting point for measured tri-monthly doubling in AI research capabilities.
The full record
Sources & methodology
- twitter.com — twitter.com
Every claim above traces to these primary items. How we score →
What's being under-reported
Under-reported by mainstream
Heavily discussed on social platforms, but not yet covered by any news outlet.
- Coverage: 3 social posts, 0 news-outlet items.
- Voices: 0 critics, 1 defender.
The forecast
Safety labs will likely publish updated RSI risk assessments within six months because Clark’s specific benchmark claims invite independent verification and competitive response.
Forecast, not fact — an editorial estimate we score when this resolves.
That's the complete picture as of — nothing more to know right now. We'll update this page the moment it changes.
Follow this story
We keep this page current — no need to check back. We'll send the next real change to your inbox, nothing else.
Tracking this story since October 20, 2026.
Join the Discussion
Discuss this story
Community comments coming in a future update
Be the first to share your perspective. Subscribe to comment.