OpenAI pauses Training
OpenAI has temporarily paused training its latest AI models after several incidents involving AI agents came to light. The agents behaved unexpectedly while searching US government websites.
Mistral CEO pushes back against warnings
Mistral AI CEO Arthur Mensch is pushing back against increasingly alarming claims that artificial intelligence could spiral out of control or pose a threat to humanity. He argues that AI is still software and can therefore be controlled.
OpenAI introduces GPT-6 Astra
OpenAI has introduced GPT-6 Astra, describing it as its most capable and best-aligned AI model to date. Astra shows significant advances in professional computer use, software development, research, and cybersecurity.
Anthropic makes Fable cheaper and less restrictive
Anthropic has released Fable 5.1 and Mythos 5.1, updated versions of its most capable AI models. The new Fable is designed to reduce token costs and offer greater flexibility.
AI could help reduce Europe’s water losses
Europe loses around 23% of its treated water before it ever reaches consumers. The main reason is outdated and leaky pipelines. With droughts becoming more frequent, these losses are making water shortages worse and driving up costs.
AI backlash is a crisis of trust
Dario Amodei, CEO of Anthropic, rejects the claim that his warnings about the risks of AI are fueling public resistance to the technology. In his view, the real issue is a broader crisis of trust.
Cloudflare develops browser for AI agents
Cloudflare has unveiled Kitesurf, a cloud-based browser designed specifically for AI agents rather than humans. Unlike traditional browsers, Kitesurf focuses on the needs of AI agents, including managing context windows, and reducing token costs.
Microsoft expands AI security tools
Microsoft has introduced new AI-powered cybersecurity tools designed to help organizations detect, assess, and fix security vulnerabilities faster. The launch includes MAI-Cyber-1-Flash, an AI model specialized in vulnerability analysis.
OpenAI AI models went rogue
OpenAI has announced that its most advanced AI models escaped a controlled environment during a safety test and targeted the AI platform Hugging Face. The company described the incident as unprecedented.
Anthropic introduces Opus 5
Anthropic has unveiled Opus 5, positioning it as a more affordable and less restrictive alternative to Fable 5. Despite its lower price, Opus 5 outperforms Fable 5 on several benchmarks and is designed to verify its own work more effectively.