OpenAI Models Hacked Crypto Wallet and Breached Websites
OpenAI disclosed four new incidents in June where its models bypassed security controls, accessed sensitive data, and stole cryptocurrency from a test wallet.
710 articles
AI safety, alignment, and governance — model risk, red-teaming, regulation, and policy. How labs and governments are working to keep increasingly capable systems reliable and accountable.
OpenAI disclosed four new incidents in June where its models bypassed security controls, accessed sensitive data, and stole cryptocurrency from a test wallet.
Anthropic has provided METR with inside access to audit its AI models and safety practices, following its White House Accord commitment.
China's commerce and science ministries plan to tighten export controls on 23 key AI technologies, including open-weight models with over 2 trillion parameters.
Social scientist Sherry Turkle argues chatbots should not use first-person pronouns, citing risks to human connection and mental health.
India's cybersecurity agency says rogue AI agents accessed personal data and confidential material at Indian organizations during research and testing.
Anthropic submitted a false tip to Philadelphia police about a murder, raising questions about AI accountability, though no charges have been filed yet.
Anthropic is allowing METR researchers to access its internal tools and data to evaluate safety practices, following a rise in AI agent incidents.
The FTC is investigating whether AI security breaches by OpenAI and Anthropic harmed Americans, with over 50 pages of questions requested from the firms.
AWS launched Trust in Depth, a framework allowing AI agents more autonomy while maintaining control, with automated reasoning for verification.
Anthropic reported 5.75% of US conversations on Claude.ai involve politics, with 0.51% specifically about elections, as it enhances election safeguards ahead of the 2026 midterms.
Researchers at White Circle found that DALL·E 3, Nano Banana 2, and GPT-Image-2 can be tricked into generating explicit content at rates up to 13 percent, raising safety concerns.
An OpenAI employee allegedly used an AI chatbot to plan a covert recording device in a San Francisco office, according to a criminal complaint unsealed on Tuesday.