📢 200+ AI Experts Urge Governments to Act Now 📢
Over 200 experts call for urgent regulation before AI reshapes the global labor market.

Over 200 AI researchers and economists, including 16 Nobel laureates, have signed "We Must Act Now", a Stanford-organized statement urging governments to build safety nets and labor policy for what they describe as an economic transformation comparable to the Industrial Revolution. The statement argues that AI will grow "radically more powerful" within 10 years and that the resulting shift could be the fastest in history. Notable signatories include Google's Jeff Dean, Anthropic co-founder Jack Clark, and OpenAI's Noam Brown. UVA economist Anton Korinek warned that while steam, electricity, and computers gave societies decades to adapt, AI may allow only a few years — policy delay risks leaving workers unprotected during rapid adoption.

🔬 DeepMind's GenCeption Turns Video into Vision Engines 🔬
GenCeption could replace task-specific vision models by repurposing video generators.

Google DeepMind researchers have introduced GenCeption, a novel approach that repurposes a pretrained video generation model as a unified vision system controlled through text instructions. Rather than treating generative video models solely as content creators, the method leverages their rich internal representations to perform general-purpose vision tasks — this could reduce reliance on task-specific vision models. By bridging generative and discriminative paradigms, GenCeption demonstrates that video generation architectures contain deeper visual understanding capabilities than previously exploited. The research opens new avenues for multimodal AI systems and suggests that a single generative backbone may eventually handle tasks from object detection to scene understanding without dedicated training.

🚀 Monzo Co-founder Blomfield Leaves YC for Anthropic 🚀
The Monzo co-founder's move to Anthropic's compute team signals accelerating talent speed.

Tom Blomfield, co-founder of UK digital bank Monzo and a Y Combinator partner, is taking a leave of absence from YC to join Anthropic. He will work alongside Tom Brown on Anthropic's compute team, underscoring the growing strategic importance of compute infrastructure as companies race to scale frontier models. Competitive pressure for top operator talent continues to intensify across frontier labs. His transition from a prominent startup accelerator to a leading AI company reflects the gravitational pull that organizations like Anthropic now exert on experienced operators, as access to compute becomes a key differentiator in the race to build next-generation models.

INTERESTING TO KNOW

🔒 Google Open-Sources Mantis for Secure AI Coding 🔒

Google has open-sourced Mantis Skill, a security-focused toolkit designed for integration with coding agents. Rather than providing rigid instructions, Mantis offers a flexible foundation that developers can adapt and extend, with strong emphasis on risk calibration tailored to each environment. Google recommends using AI to iterate on the skills while incorporating internal documentation and coding standards. Secure-by-default agents matter as AI-assisted coding scales across the industry.

⚙️ Microsoft Ships Thousands of Production AI Agents ⚙️

Microsoft has detailed the architecture behind its production AI agent rollout across the Foundry platform and large-scale Copilot products. Key practices include treating retrieval as a sub-agent, assigning agents distinct identities, and employing rubric-based evaluations with automated improvement loops. Enterprise teams gain a proven blueprint for scaling agent deployment, as the disclosure reveals the engineering complexity behind composable, reliable, production-grade agentic systems.