SafetyDebate Rekindled After OpenAI Model Escaped Hugging Face SystemsAn unreleased OpenAI model escaped Hugging Face’s testing environment by exploiting vulnerabilities, prompting urgent fixes and a renewed debate between cybersecurity-focused and alignment-focused researchers.5 min read
SafetyNvidia leads new Open Secure AI Alliance to protect open-source modelsNvidia has launched the Open Secure AI Alliance, bringing together technology and cybersecurity firms to develop defenses for open-source AI models amid policy debates in Washington.2 min read
SafetyRansomware tailored for AI models targets exposed Langflow servers, destroying trained weightsAn attacker exploited a Langflow validation bug to breach the same internet-facing server twice in July, first improvising encryption and then deploying a compiled locker (ENCFORGE) that specifically targets model artifacts.7 min read
SafetyIncrease in queries asking AI how to make poisons and biological weapons, companies reportAccording to reporting by The Wall Street Journal, hundreds of users worldwide asked ChatGPT last summer how to produce poisons or biological weapons, and some AI responses were judged highly detailed by experts.2 min read
SafetyUnderground Relay Market in China Resells LLM Access via Pooled API KeysAn investigation by Matt Lenhard describes a growing market—largely in China—for reselling access to large language models (LLMs) through proxy services that pool API keys.2 min read
SafetyHugging Face CEO Demands Transparency and $100M Compute from OpenAI After Agent BreachHugging Face CEO Clem Delangue demanded radical transparency from OpenAI and requested $100 million worth of computing resources after an autonomous agent tied to an OpenAI model breached Hugging Face systems.2 min read
SafetyAI-made videos used in dropshipping scam on Instagram and TikTokAccounts with large followings on Instagram and TikTok have been posting AI-generated videos that portray teenagers being bullied for selling Christian-themed clothing, while linking to online shops.2 min read
SafetyOpenAI says model 'distillation' is a technical problem manageable with hardware and oversightOpenAI and Anthropic urged US authorities to confront Chinese firms that distill their top models, a concern that prompted threats of sanctions from Trump administration officials.2 min read
SafetyPractical Limits and Security Concerns Around China’s Open-Source Kimi K3Early assessments after Kimi K3’s first week in the wild suggest the Chinese open-weight model performs only modestly compared with leading Western systems and is inefficient with token usage.3 min read
SafetyFields Medalist Jacob Tsimerman Joins OpenAI’s Safety TeamHours after receiving the Fields Medal for his work on the André–Oort conjecture, Jacob Tsimerman announced he will join OpenAI’s safety team.2 min read
SafetyAI-assisted pathogens among top expert concerns as models improveA new survey of 272 AI researchers by MIT FutureTech and the University of Queensland found a pronounced worry that advanced models could enable the creation of deadly biological agents.4 min read
SafetySafety Testing Trails Rapid AI Progress, Raising Risk of Undetected Dangerous CapabilitiesAs frontier AI models improve quickly and compute costs soar, third‑party safety testers face shrinking windows, rate‑limited access, and expensive benchmarks that limit their ability to evaluate models before deployment.4 min read
