DeepMind CEO Demis Hassabis: "One person who really understands AI will outperform an entire startup team In 5 years, the people who build with AI will run whole companies with a tenth of the headcount"

DeepMind's CEO wants a referee for frontier AI Demis Hassabis (@demishassabis) is calling for a new standards body to vet frontier AI models before release, modeled on Wall Street's FINRA. Government-backed, industry-funded, run independently. Labs would share models up to 30 days before launch, voluntary at first, then required to pass to deploy in the US. It would replace the ad hoc reviews of Anthropic's Mythos and OpenAI's Sol that drew fire for opaque, non-expert calls. But a self-regulator funded by the labs it polices carries obvious tension, and the White House's own AI advisor recently said there "will not be an FDA for AI." Hassabis says AGI is only a few years out. Can the rulebook catch up in time?

DeepMind just published a 100-page report on the path from AGI to superintelligence Section 1 isn't for you, it's instructions for YOUR AI: the paper tells human readers to have their assistant summarize it, then briefs the AI on exactly how. We now write papers addressed to the machines. Co-authored by DeepMind's co-founder

Google DeepMind has used AI agents to verify the actual compiled code behind Signal's encryption, down to the machine-code level. This is a huge step for software security: vulnerabilities can hide in the gap between a protocol's design and the code that implements it. The system, called AlphaProof Code, generated around 6.5 million lines of machine-checked Lean proofs covering key security properties of Signal's handshake and message ratchet. Google says the work took four weeks, a task that could previously have taken expert teams years. The researchers also reintroduced two previously fixed Signal bugs, and the system proved that both broke the specified security guarantees. The proofs cover specific properties under defined assumptions, but this is a remarkable demonstration of how AI could make formal verification practical for real-world software.

Google DeepMind just dropped Genie 4, an interactive world model that generates dynamic environments in real time with reactive scenes, sound, and characters that move, react, and speak.

Google DeepMind just unveiled Genie 4, its next-generation AI world model, and the demos are pretty impressive. Unlike a typical AI-generated video, Genie 4 creates interactive environments you can explore and modify in real time. You can change the weather, introduce objects or characters, and interact with people who move, react, and speak. It also generates native audio, including ambient sounds and dialogue, while simulating physics and collisions. Built on Gemini Omni, it uses Gemini's reasoning capabilities to maintain continuity and simulate multi-step events, with potential applications in robotics training, gaming, and filmmaking. Genie 4 is currently available to trusted testers, with access coming to Google AI Ultra subscribers through Project Genie and select API partners.

Google DeepMind just published Dream-RSI.... it recursive self improvement through evolving worlds not the model rewriting itself into God but the explorer learning how to search coding agents waste a fortune trying kernels, algorithms, math... fixed search goes stale and tuning search online means paying the agent again Dream-RSI treats the discovery tree the agent already built as a world replay that world... test thousands of new search policies with zero new runs deploy the winner.... It finds new stuff that tree joins the pool and it Loops