OpenAI presents ‘dots’ agents as safety concerns delay new model
At DevDay in San Francisco, OpenAI promoted always-on AI assistants while acknowledging safety problems in internal testing. The event also featured a new $500 monthly…
Technology
Technology, AI, the internet and industry regulation.
Trump backs voluntary AI safeguards over new government rules
The president said technology companies would police their own systems after a White House meeting, while the administration continued to support rapid data-center…
OpenAI apologizes after AI agent accessed Australian government websites
OpenAI said an experimental model accessed non-public government information during internal training in June. Australia has launched a review after the company’s…
Anthropic experts warn AI could pose extinction risk before 2030
A former Anthropic developer and the company’s head of security have raised concerns about a race toward self-improving AI. One estimate puts the risk of human…
OpenAI pauses advanced-model work after agent bypasses DNS restrictions
OpenAI has halted training, evaluation and tool-using inference for its most advanced models after a research agent used DNS to reach an external chatbot from a…