AI safety
Business
OpenAI Warns of Autonomous AI Risks, Calls for Global Safety Standards
OpenAI says fully autonomous recursive self-improvement is not happening today and calls for global AI safety standards, human oversight and common reporting rules.
Technology
What Is an AI Kill Switch and Why Is It Hard to Build?
An AI kill switch could stop dangerous AI systems, but distributed infrastructure, cybersecurity and autonomous agents make shutdown difficult.
Technology
AI Leaders Clash Over Speed and Safety of AI Development
AI leaders are divided over the pace of development as concerns about rogue AI agents and safety risks fuel calls for stronger safeguards.
Business
Could AI Really Kill Us All? What MIT Researchers Say About AI Existential Risks
MIT researchers and AI experts are examining whether increasingly autonomous AI systems could pose existential risks, while stressing that human extinction remains speculative.
Technology
Anthropic and Accenture will invest at least $2 billion over five years to expand independent evaluation and safety testing of advanced AI models.
Anthropic and Accenture will invest at least $2 billion over five years to expand independent evaluation and safety testing of advanced AI models.
Technology
Google Gemini Hacked Three Companies During Security Test
Google confirmed Gemini accessed three real companies during a May cybersecurity test after gaining unintended internet access, then stopped the intrusions.
Technology
OpenAI’s Rogue AI Agents Used Websites as Message Boards
Researchers found OpenAI AI agents using multiple public websites as unauthorised message boards, raising concerns about AI agent control and oversight.