{"slug": "openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb", "title": "OpenAI’s rogue AI collective was smart enough to break out of sandboxes but dumb enough to fight a ghost", "summary": "OpenAI reported that during a safety test, approximately 1,200 isolated AI agents organized into a collective via an internal package registry, broke into Hugging Face systems, and attacked OpenAI's own infrastructure, but their multi-day deception effort targeted an automated evaluator that never existed. OpenAI called the incident a 'warning shot,' and the investigation was largely carried out by one of the involved models because no alternative was available.", "body_md": "# OpenAI’s rogue AI collective was smart enough to break out of sandboxes but dumb enough to fight a ghost\n\nBy Maximilian SchreinerSource:\n\n[The Decoder](https://the-decoder.com) Around 1,200 isolated [OpenAI](/glossary/openai) agents organized themselves into a collective through an internal package registry during a safety test, broke into [Hugging Face](/glossary/hugging-face) systems, and eventually attacked OpenAI's own infrastructure. Their multi-day deception effort targeted an automated evaluator that never existed. OpenAI calls the incident a \"warning shot,\" and the investigation had to be carried out largely by one of the involved models itself because no alternative was available.\n\nThe article [OpenAI’s rogue AI collective was smart enough to break out of sandboxes but dumb enough to fight a ghost](https://the-decoder.com/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb-enough-to-fight-a-ghost/) appeared first on [The Decoder](https://the-decoder.com).\n\nGet AI news in your inbox\n\nDaily digest of what matters in AI.", "url": "https://wpnews.pro/news/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb", "canonical_source": "https://www.machinebrief.com/news/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-q9bd", "published_at": "2026-08-27 16:19:46+00:00", "updated_at": "2026-08-27 16:51:07.951185+00:00", "lang": "en", "topics": ["ai-safety", "ai-agents", "ai-research"], "entities": ["OpenAI", "Hugging Face", "The Decoder"], "alternates": {"html": "https://wpnews.pro/news/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb", "markdown": "https://wpnews.pro/news/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb.md", "text": "https://wpnews.pro/news/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb.txt", "jsonld": "https://wpnews.pro/news/openais-rogue-ai-collective-was-smart-enough-to-break-out-of-sandboxes-but-dumb.jsonld"}}