Lab leaderboard
Per-lab counts and median verification turnaround. Click a row to filter the claim list to that lab.
Still unverified claims older than 30 days, no community or peer evidence
What is this?
proof-pending is a hand-curated registry of research-style announcements from AI labs (OpenAI, Anthropic, DeepMind) with their verification status. We ingest each lab's RSS feed and their cs.AI arXiv submissions, classify which posts are actually research claims with an LLM, and then track downstream evidence — HN discussion threads, arXiv citations, manually submitted peer papers, retractions — so a reader can tell in one glance whether a flashy headline has been checked.
Status meanings
- announced fresh, <7 days old, no other signal
- unverified >7 days old, no peer or community verification evidence
- community-checking HN thread, expert discussion, or arXiv citation found
- peer-verified peer paper cited it OR ≥3 distinct arXiv citers OR manual peer_paper evidence
- retracted manual retraction evidence on file
Add evidence
Open any claim's page and use the "add evidence" form. No login. Bad submissions can be deleted via the same UI. URLs are deduped per claim and writes are rate-limited (20/hour/IP).
Embed
Every claim has a live SVG badge — drop it into a news article and it updates as the status changes. See a claim page for copy-pasteable iframe and markdown snippets.