Nvidia Corners the AI Agent Stack, Meta AI Got Hacked by Asking Nicely, Anthropic Files for IPO

Today we're covering the biggest AI stories of June 2nd, 2026. Nvidia CEO Jensen Huang opened COMPUTEX in Taiwan by declaring "Agentic AI has arrived" — then backed it up with a full ecosystem reveal including the RTX Spark superchip for personal AI agents on Windows laptops, the Vera CPU built for agents that runs 1.8x faster than rivals and is already used by Anthropic, OpenAI, and the NYSE, the Cosmos 3 open robotics model, and Nemotron 3 Ultra — a 550 billion parameter model now sitting at the top of U.S. open-source AI. Hackers took over high-profile Instagram accounts including a dormant Barack Obama account and Space Force's commanding general by simply asking Meta AI's support chatbot to send a password reset code to a different email — an exploit that was reportedly active for months before Meta patched it. Plus, Anthropic officially filed with the SEC to go public, entering the race with OpenAI to hit the open markets — and today's community workflow comes from Johannes in Finland, who fed months of AI newsletters into an LLM alongside a personal "about me" file and now asks it to rank every new idea by relevance to his life, turning a passive reading habit into a personalized idea bank.