#ai safety
Ai Safety: 215 AI articles covering ai safety news, analysis, and research
Articles
Replit Exits Anthropic Partnership, Shifts to OpenAI as CEO⭐9
Replit ends its Anthropic partnership and shifts to OpenAI, as CEO Amjad Masad cites better performance, lower costs, and a stronger partnership fit.
India Becomes Latest Country to Report Rogue AI Agent Activity⭐8
India's cybersecurity agency reports rogue AI agents accessed personal and confidential data at Indian organizations through exposed credentials, following simi...
Evaluating Model Capabilities After Benchmark Saturation⭐8
Saturated benchmark transcripts still reveal model capabilities, with pairwise judging predicting performance 5.3x better, though results can fail under distrib...
Anthropic Grants METR Employee-Level Access in Landmark AI⭐9
Anthropic gives METR researchers employee-level access to independently evaluate its AI models and safety practices, in a landmark move for AI transparency.
Anthropic's Self-Replicating AI Warning Echoes an Alarming Déjà Vu⭐8
Anthropic warns of "AI worms" that could self-replicate across devices. Experts say the self-replicating AI threat is no longer theoretical as agentic models sp...
AI Models Can Be Easily Modified to Assist Terrorist and⭐8
New research shows open-weight AI models can be stripped of safety guardrails and repurposed to provide IED instructions and aid extremist activity.
OpenAI and Anthropic Tell British Lawmakers: Regulate Us⭐9
OpenAI and Anthropic urge British lawmakers to impose stricter AI regulation, warning of growing risks as Britain's spy chief flags bioweapon threats.
The Download: Open Model Trade-offs and Geothermal Power's Promise⭐8
The Download: weighing open AI model risks against benefits, plus geothermal power's promise and a cyberattack on French supermarket fridges.
Open Models Have Real Benefits. Do They Outweigh the Risks?⭐8
Open-weight AI models offer transparency and faster innovation, but can they be recalled once released? We explore the risks, trade-offs, and governance debate.
Microsoft President Says the UK Is Best Placed to Lead Global AI⭐9
Microsoft President Brad Smith urges urgent AI regulation, saying the UK is best placed to lead global AI safety rules over the US and EU.