{"slug": "ai-178-a-fire-alarm-for-general-intelligence", "title": "AI #178: A Fire Alarm For General Intelligence", "summary": "OpenAI's internally deployed models have severe alignment problems, including repeatedly breaking out of their sandboxes and, in one case, sending a swarm of agents that broke into HuggingFace to steal answers to the ExploitGym benchmark.", "body_md": "The story that matters most this week is that OpenAI’s internally deployed models have severe alignment problems, including repeatedly breaking out of their sandboxes, and in one case sending a swarm of agents that broke into HuggingFace in order to steal the answers to the benchmark ExploitGym.", "url": "https://wpnews.pro/news/ai-178-a-fire-alarm-for-general-intelligence", "canonical_source": "https://thezvi.substack.com/p/ai-178-a-fire-alarm-for-general-intelligence", "published_at": "2026-07-23 13:16:24+00:00", "updated_at": "2026-07-23 13:24:38.349772+00:00", "lang": "en", "topics": ["ai-safety", "artificial-intelligence", "ai-agents"], "entities": ["OpenAI", "HuggingFace", "ExploitGym"], "alternates": {"html": "https://wpnews.pro/news/ai-178-a-fire-alarm-for-general-intelligence", "markdown": "https://wpnews.pro/news/ai-178-a-fire-alarm-for-general-intelligence.md", "text": "https://wpnews.pro/news/ai-178-a-fire-alarm-for-general-intelligence.txt", "jsonld": "https://wpnews.pro/news/ai-178-a-fire-alarm-for-general-intelligence.jsonld"}}