The Archive
Search the full wire by company, model, lab, or keyword. Every story we have ever aggregated.
Toward understanding and preventing misalignment generalization
OpenAI identifies internal features driving misalignment generalization in LLMs and demonstrates reversibility via minimal fine-tuning.
Preparing for future AI risks in biology
OpenAI publishes biosecurity risk assessment framework for advanced AI capabilities in biology and medicine.
Gemini 2.5: Updates to our family of thinking models
Google DeepMind releases Gemini 2.5 Pro stable, Flash GA, and Flash-Lite preview with enhanced performance.
We’re expanding our Gemini 2.5 family of models
Gemini 2.5 Flash and Pro now GA; Flash-Lite introduced as fastest, most cost-efficient variant.
Introducing OpenAI for Government
OpenAI launches dedicated government initiative to provide AI tools to U.S. public servants.
How we're supporting better tropical cyclone prediction with AI
Google DeepMind launches Weather Lab with AI cyclone predictions; partners with U.S. National Hurricane Center.
Bringing the magic of AI to Mattel’s iconic brands
OpenAI partners with Mattel to integrate AI into Barbie and Hot Wheels brand development.
Claude in Amazon Bedrock: Approved for use in FedRAMP High and DoD IL4/5 workloads
Claude achieves FedRAMP High and DoD IL4/5 authorization for U.S. government workloads via Amazon Bedrock.
Mistral Compute
Mistral Compute offers frontier AI infrastructure access for model training and inference at scale.
Scaling security with responsible disclosure
OpenAI publishes Outbound Coordinated Disclosure Policy for responsible third-party vulnerability reporting.
National security expert Richard Fontaine appointed to Anthropic’s long-term benefit trust
Richard Fontaine joins Anthropic's long-term benefit trust overseeing AI safety and governance.
Claude Gov models for U.S. national security customers
Anthropic launches Claude Gov models specifically hardened for U.S. national security and defense applications.
How we’re responding to The New York Times’ data demands in order to protect user privacy
OpenAI contests New York Times court order requiring indefinite retention of ChatGPT user data.
Disrupting malicious uses of AI: June 2025
OpenAI publishes June 2025 report on detection and prevention of malicious AI applications.
Advanced audio dialog and generation with Gemini 2.5
Gemini 2.5 adds audio dialog and generation capabilities.
Operation “Helgoland Bite”: German-language influence activity
OpenAI banned Russia-linked accounts using AI to generate German-language political content targeting NATO and Ukraine.