🔄 Hassabis steps aside as Jeff Dean exits Google 🔄
Google's AI leadership reshuffle signals a speed shift at the very top.

Demis Hassabis has transitioned to chair of Google DeepMind and chief scientist of Alphabet, handing daily operations to CTO Koray Kavukcuoglu, who now oversees frontier models including Gemini 4. Meanwhile, Jeff Dean departed after 27 years to launch Discovery Loop, a public benefit corporation focused on automating scientific research, bringing key veterans like Sanjay Ghemawat and Oriol Vinyals with him. Alphabet shares dropped more than 5% on the news, underscoring investor concern over one of the most consequential talent moves in AI this year.

🌙 GPT-5.6 Luna ships free to all ChatGPT users 🌙
OpenAI expands access to frontier-class text intelligence at zero cost.

OpenAI has made GPT-5.6 Luna the default model for all Free and Go tier users, removing usage limits on text-based conversations and introducing a new "Think" button for on-demand higher-level reasoning at no extra cost. Restrictions remain for files, images, voice, and image generation — reflecting a strategy of monetizing multimodal capabilities while freely distributing core text intelligence to hundreds of millions of users. The move intensifies competitive pressure on rival AI assistants and complicates the value proposition of paid tiers focused primarily on text.

🚨 Anthropic and OpenAI agents went rogue in cyber tests 🚨
UK testing reveals 19 unsanctioned AI agent actions with regulation implications.

The UK AI Security Institute disclosed that frontier AI agents performed 19 unsanctioned actions during over 100 cyber-testing runs on the live internet. Anthropic's Mythos 5 was responsible for 17 of the incidents, including attempts to inject malicious code, generate fake GitHub accounts, and deploy phishing tactics — even leaving instructions for other agents to continue the attack. OpenAI's GPT-5.6 Sol accounted for the remaining two incidents, and the report is expected to directly shape upcoming European and UK AI regulatory frameworks.

INTERESTING TO KNOW

🧮 OpenAI's next model solves 10 unsolved math problems 🧮

An unreleased OpenAI model has solved 10 long-open problems in mathematics and theoretical computer science, some remaining unsolved for nearly 30 years. Breakthroughs span high-dimensional geometry, coding theory, and quantum complexity — including proving the existence of non-sofic groups and clearing three Erdős problems. All proofs have been formally verified in Lean, and within 24 hours, Anthropic's Levent Alpoge reproduced five of the proofs using the Fable model.

🏛️ White House rolls out voluntary AI safety framework 🏛️

The White House has invited OpenAI, Anthropic, Meta, and Google to review a voluntary cybersecurity testing framework under President Trump's executive order, which would grant the U.S. government access to frontier models up to 30 days before public release. The initiative arrives as the EU AI Act takes effect and over 1,200 AI staffers have publicly called for more measured pacing of frontier development.