Boko Haram Used AI Chatbots to Plan Motorcycle-Jump Attack
The terrorist group used chatbots to modify motorcycles for jumping defensive trenches, plus bomb-building and attack planning, per 60 interviews with defectors.
Grok AI Reportedly Uploads User Directory to xAI Servers
A researcher discovered xAI's Grok Build CLI uploaded entire Git repositories, including unread and sensitive files like credentials and full commit history, to xAI's Google Cloud storage. This occurred even with the "Improve the model" setting disabled.
New York Pauses Approval of Large AI Data Centers
New York paused permits for new large-scale data centers for up to 12 months while developing environmental and grid-impact standards, reflecting growing regulatory scrutiny of AI infrastructure expansion.
DeepMind Proposes Independent U.S. Body for Frontier AI Safety Reviews
DeepMind CEO Demis Hassabis proposed an independent U.S. standards body to evaluate frontier AI models before release, including testing for cyber, bio, and deception risks.
OpenAI and Anthropic Warn of AI Model Distillation by Chinese Firms
OpenAI and Anthropic warned U.S. policymakers that Chinese AI companies, including Alibaba and DeepSeek, are using model distillation to replicate frontier AI capabilities, calling for stronger protections against unauthorized capability transfer.
Nobel Laureates, AI Leaders Urge Action on Economic Impact
Over 200 economists and AI researchers, including Nobel laureates and AI company leaders, issued a joint statement coordinated by Stanford Digital Economy Lab, calling for immediate action to prepare for AI's potentially unprecedented economic transformation and large-scale job displacement.
AI Futures Project Releases Long-Term Scenarios for AI Governance
The AI Futures Project released AI 2040, presenting five scenarios for AI development through 2040 and recommending policies such as greater transparency, international coordination, and frontier AI governance.