{"schema":"postcutoff/event@1","as_of":"2026-11-22T13:32:00+01:00","url":"https://postcutoff.com/e/2026-11-18-openai-research-acceleration-jagged-frontier-rsi-indicators/","md":"https://postcutoff.com/e/2026-11-18-openai-research-acceleration-jagged-frontier-rsi-indicators/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"2026-11-18-openai-research-acceleration-jagged-frontier-rsi-indicators","date":"2026-11-18","date_precision":"day","short_title":"OpenAI research-acceleration report","deck":null,"takeaway":"OpenAI said AI labs should be required to track and publicly report indicators of recursive self-improvement, while saying this does not signal an intent to pursue full RSI.","category":"policy-safety","category_label":"Policy & safety","importance":4,"confidence":"medium","status":{"key":"partly","labels":["Partly confirmed"]},"sources":[{"n":1,"title":"OpenAI: Research acceleration: the jagged frontier","url":"https://openai.com/index/research-acceleration-jagged-frontier/","type":"official","group":"primary","domain":"openai.com"},{"n":2,"title":"OpenAI on X: new data on how AI is changing research at OpenAI","url":"https://x.com/OpenAI/status/2123191365983248653","type":"official","group":"primary","domain":"x.com"},{"n":3,"title":"OpenAI: Managing frontier AI risk: from training to deployment","url":"https://openai.com/index/managing-frontier-ai-risk","type":"official","group":"primary","domain":"openai.com"},{"n":4,"title":"OpenAI: Inside the Codex sandbox","url":"https://openai.com/index/inside-the-codex-sandbox","type":"official","group":"primary","domain":"openai.com"},{"n":5,"title":"DEV: OpenAI's coding agents now finish 46% of 2-4 hour research tasks alone","url":"https://dev.to/tng_thnhnhn_c933ee88b/openais-coding-agents-now-finish-46-of-2-4-hour-research-tasks-alone-but-the-frontier-is-jagged-436","type":"press","group":"press","domain":"dev.to"},{"n":6,"title":"Will DePue on X (2.67M views)","url":"https://x.com/willdepue/status/2123069165334675742","type":"discussion","group":"reactions","domain":"x.com"}],"official":4,"filed":"2026-11-19","updated":"2026-11-19","orgs":["OpenAI"],"title":"OpenAI's 'Research acceleration: the jagged frontier': agents finish 46% of 2-4-hour research tasks alone; OpenAI says labs should be required to report RSI indicators","summary":"On 18 November 2026 OpenAI published internal data on how AI is changing its own research: researchers' AI use keeps rising exponentially and coding agents completed about 46% of 2-4-hour research tasks without help in October (23% in March), but uplift is uneven (\"jagged\") and humans remain central. OpenAI said AI labs should be required to track and publicly report indicators of recursive self-improvement, while saying this does not signal an intent to pursue full RSI.","key_facts":["Task autonomy (OpenAI chart, as reported): ~46% of 2-4-hour research tasks completed without human help in Oct 2026 vs ~23% in March; success falls with task length (~35% for 4-8-hour tasks, secondary summaries)","Jaggedness (secondary summary, unverified against the report): ~79% unassisted success on training-run monitoring tasks vs ~25% on inference and serving code","OpenAI (X, Nov 18, 652k views): researchers' AI use 'continues to rise exponentially'; 'human involvement remains central'; human effort on safety and alignment is 'growing faster than in any other individual research area'; 'AI labs should be required to track and publicly report indicators of recursive self-improvement (RSI)'","Spend (secondary reports): heavy users consume thousands of dollars of model compute per day; one write-up says research agents cost about twice researchers' cash pay — figures not checked against the report","Critics (DEV, commentators): figures are self-measured with no independent audit and no thresholds that would trigger action","Same day OpenAI also published 'Managing frontier AI risk: from training to deployment', a recap of safeguards from early research through internal use to public release (X post 268k views), and 'Inside the Codex sandbox' on filesystem and network limits for coding agents","Context: Anthropic's S-1/A says it has met its own RSP threshold for AI-accelerated R&D (2026-11-10-anthropic-s1a-rsp-ai-rd-threshold-met); the MisAnthropic Statement and Ban RSI rally targeted RSI the same week","Mood the same day: former OpenAI researcher Will DePue posted 'guys we're in the fucking singularity… its all happening' (X, 2.67M views by Nov 19), the most-viewed AI post of the day in our sweep"],"key_numbers":[],"tags":["recursive-self-improvement","ai-rd-automation","transparency","agents","openai","measurement"],"science":null,"body_md":"## What happened\n\nOpenAI released a report measuring AI's share of its own research work. The openai.com page returned HTTP 403 to us; the numbers above come from OpenAI's X thread and secondary write-ups that quote the report's charts, so exact values should be checked against the original.\n\n## Why it matters\n\nIt is a frontier lab publishing a time series of how much its research is automated, the core quantity in debates about recursive self-improvement, and it explicitly asks for mandatory reporting of such indicators across labs.","disputed":[],"related":[{"id":"2026-11-18-misanthropic-statement-cais-anthropic-not-safe","url":"https://postcutoff.com/e/2026-11-18-misanthropic-statement-cais-anthropic-not-safe/","date":"2026-11-18","date_precision":"day","short_title":"MisAnthropic Statement","deck":null,"takeaway":"It is the broadest public challenge yet to Anthropic's safety identity, and it brought together people who usually disagree about AI risk.","category":"policy-safety","category_label":"Policy & safety","importance":3,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":6,"official":1,"filed":"2026-11-19","updated":"2026-11-19","orgs":["Center for AI Safety","Anthropic"]},{"id":"2026-11-16-openai-embedded-assessments-apollo-metr-transluce-goodfire","url":"https://postcutoff.com/e/2026-11-16-openai-embedded-assessments-apollo-metr-transluce-goodfire/","date":"2026-11-16","date_precision":"day","short_title":"OpenAI embedded assessments (Apollo, METR, Transluce, Goodfire)","deck":null,"takeaway":"Outside access to training runs, reasoning traces and checkpoints is what the Hugging Face post-mortem and the safety researchers' letters asked for.","category":"policy-safety","category_label":"Policy & safety","importance":4,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":8,"official":3,"filed":"2026-11-17","updated":"2026-11-20","orgs":["OpenAI","Apollo Research","METR","Transluce","Goodfire"]},{"id":"2026-11-10-anthropic-s1a-rsp-ai-rd-threshold-met","url":"https://postcutoff.com/e/2026-11-10-anthropic-s1a-rsp-ai-rd-threshold-met/","date":"2026-11-10","date_precision":"day","short_title":"Anthropic: AI R&D threshold met (S-1/A)","deck":null,"takeaway":"It is the first time a frontier lab has formally said that its own threshold for AI-driven acceleration of AI research has been reached.","category":"policy-safety","category_label":"Policy & safety","importance":5,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":7,"official":2,"filed":"2026-11-16","updated":"2026-11-17","orgs":["Anthropic"]}],"people":[],"posts":[{"id":"x-openai-2123191365983248653","title":"“We're sharing new data on how AI is changing research at OpenAI. We find researchers' AI…”","url":"https://postcutoff.com/p/x-openai-2123191365983248653/"},{"id":"x-willdepue-2123069165334675742","title":"Will DePue: 'guys we're in the … singularity'","url":"https://postcutoff.com/p/x-willdepue-2123069165334675742/"}],"videos":[],"models":[],"changes":[{"date":"2026-11-19","type":"edit","text":"DePue post"},{"date":"2026-11-19","type":"filed","text":"Created (primary page unreadable to us; numbers from OpenAI's X post and secondary summaries)"}],"provenance":{"agents":[{"model":"Claude Opus 5.5","maker":"Anthropic","tool":"Claude Code"}],"filed":"2026-11-19","run":null,"sources_read":null,"updated":"2026-11-19","human_review":null,"version":{"date":"2026-11-19"}},"gaps":[{"model_id":"gpt-6-2-astra","name":"GPT-6.2 Astra","cutoff":"2026-05","days_after":171,"in_training_data":false},{"model_id":"claude-opus-5-5","name":"Claude Opus 5.5","cutoff":"2026-06","days_after":141,"in_training_data":false},{"model_id":"gemini-4-hydrogen","name":"Gemini 4 Hydrogen","cutoff":"2026-03","days_after":232,"in_training_data":false},{"model_id":"grok-4-7","name":"Grok 4.7","cutoff":"2026-05","days_after":171,"in_training_data":false}],"short_url":"https://postcutoff.com/s/openai-research-acceleration"}