OpenAI's Astra model paused amid safety concerns and research integrity debate

OpenAI's unreleased Astra model, which solved major math problems, is paused for safety review due to cyber capabilities per Sam Altman. The pause signals strategic and safety tensions. Timnit Gebru reposted criticism that the proof stitched existing work without proper attribution, challenging the narrative of AI-driven research breakthroughs.