OpenAI's Astra math 'breakthroughs' commit research misconduct, mathematicians say
asusarla · x · 2026-09-07
OpenAI claimed its next-gen LLM Astra resolved or progressed 10 long-standing open math problems during internal testing, at a token cost of just $2,000, publishing a 250-page paper. But experts who reviewed it found two of the most exciting results reuse existing ideas from recent mathematical literature without proper citation, contradicting the press release's claim that the problems had seen no progress.
Per Scientific American, mathematicians argue this constitutes research misconduct, highlighting how AI-generated results are being released faster than human academic rigor can validate them. Timnit Gebru amplified the story, contrasting its visibility with OpenAI's original hype.
Related event: Mathematicians Accuse OpenAI's Astra Math Results of Research Misconduct(2 posts)→
More from Models
- Next-gen model training could hit 5x pretraining compute if 300k GB200 rumor holds — scaling01 · 2026-09-07
- GPT-6 could use 2-5x pre-training compute for RL, speculates scaling01 — scaling01 · 2026-09-07
- Banned triton is reward hacking; unbanned PyPI shortcut is a task bug — xeophon · 2026-09-07
- Frontier pre-training runs likely capped at ~2 months to avoid wasting algorithmic progress — scaling01 · 2026-09-07
- Codex recreates a full game in 40 minutes from one vague prompt, no Blender skills needed — FuSheng_0306 · 2026-09-07
- Vincent Conitzer shows what GPT-6 Astra can't do, pushing back on loose AGI labels — conitzer · 2026-09-07