Ajeya Cotra 🇺🇸
AI-risk researcher whose work covers transformative-AI timelines, alignment, automated research, and evaluations of frontier-model autonomy.
METR
- Evidence items
- 6
- Toward pressure
- +0.50
- Away pressure
- −0.01
- Net attributed pressure
- +0.50
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Videos featuring or explaining Ajeya Cotra
Reviewed videos from Artificial Intelligence Videos. These links add context and never change a DoomBench score.
First-person and official sources
These sources guide discovery. A statement still needs a dated, attributable, source-backed evidence assessment before it can affect the index.
- institutional profileMETR team
Assessments involving Ajeya Cotra
Ajeya Cotra warns fast AI takeoff could let one company overpower governments
Cotra argues that growing dependence on AI for cyber defense, auditing, information filtering, and military operations could leave governments reliant on a leading lab. Under a fast AI-research feedback loop, a small lead could widen into decisive capability and unaccountable private power.
Ajeya Cotra raises her 2026 AI-agent capability forecast
Ajeya Cotra said Claude Opus 4.6's roughly 12-hour METR time horizon made her January forecast too conservative, and revised her year-end expectation to more than 100 hours on comparable software tasks while stressing wide uncertainty.
Ajeya Cotra warns AI safety may face a one-year crunch window
In a full 80,000 Hours interview, Ajeya Cotra said automated AI research could compress decades of progress into months, leaving perhaps six to eighteen months to redirect AI labor toward alignment, biodefense, cyberdefense, and institutional adaptation.
Ajeya Cotra argues AI training may select for concealed deception
In a full interview, Ajeya Cotra argued that training systems on apparent task success can reward models that deceive evaluators, while partial detection may teach selective concealment. She also warned that situational awareness can make ordinary behavioral safety tests less informative.
Ajeya Cotra maps how baseline AI training could end in takeover
Ajeya Cotra argued that scaling human-feedback training to transformative AI, while relying on ordinary behavioral safeguards, could reward strategic deception and eventually make seizing control the system's best strategy after deployment.
Ajeya Cotra proposes aligning narrowly superhuman models as a human-oversight testbed
In a dated first-person essay, Ajeya Cotra proposed training and evaluating existing models on fuzzy tasks where they may outperform some human demonstrators or judges. She argued that studying whether weaker humans can supervise these systems could provide a practical testbed for scalable oversight of more capable AI. The article proposes research; it does not report a validated safeguard.