Welcome to the AI Daily Briefing for Tuesday, June 2nd, 2026. I'm Bob, and here's what happened in AI over the last 24 hours.

GPT-5.6 rumors are reaching a fever pitch, and this time there might actually be fire behind the smoke. Multiple insiders are pointing to a launch this week, potentially at an OpenAI "Intelligence at Work" livestream with Sam Altman. The leaks describe a model that reaches what some are calling "Mythos-level" performance — matching or rivaling upcoming Anthropic models — but at two to three times lower pricing. We're hearing about a one point five million token context window, major gains in reasoning, frontend code generation, personality, and agentic workflows. One insider said it could have easily been called GPT-6. Meanwhile, rumors suggest Anthropic is also close to releasing Mythos-class models to all customers, currently restricted to cybersecurity partners. June is shaping up to be an arms race.

NVIDIA's Computex keynote in Taipei set the agenda for AI hardware in 2026. Jensen Huang unveiled the Vera Rubin NVL72 architecture — the next generation of AI data center infrastructure claiming ten times inference performance per watt. But the real headline grabber is the RTX Spark Superchip we covered yesterday, now with more detail: twenty Arm cores, six thousand CUDA cores, a hundred twenty-eight gigs of unified LPDDR5X memory, capable of running one hundred twenty billion parameter LLMs locally with million-token context windows. Huang called it bigger than the smartphone shift. Microsoft, Dell, HP, Lenovo, ASUS, and Acer are all building RTX Spark systems for fall 2026. NVIDIA also announced Jetson Thor for physical AI and robotics, and a strengthened partnership with Marvell for AI networking that sent Marvell shares soaring.

Google and Meta jointly released PaGeR — a new AI model for three hundred sixty degree scene geometry. It's state of the art on panoramic depth estimation, metric depth, surface normals, and sky masks. Code and project page are public. This is the kind of foundational research that makes augmented reality and spatial computing more practical, and it's notable that Google and Meta collaborated on it.

Florida is suing OpenAI. Attorney General James Uthmeier accuses the company of promoting ChatGPT despite its alleged links to, quote, "self-harm, cognitive decline, and behavioral addiction." The state is seeking penalties and a court order rather than criminal charges, but a separate criminal investigation into OpenAI remains ongoing. This is the most aggressive regulatory action against an AI company from a US state so far, and it signals that the legal landscape for consumer AI is about to get a lot more complicated.

OpenAI Codex continues its rapid expansion. Computer use is now available on Windows, letting Codex see your screen and perform tasks on your device — it was previously Mac only. Version zero point one thirty six shipped with clickable links and tables in the terminal, archive sessions, resume threads with MCP status, and sandbox hardening. You can also manage and review Codex jobs remotely from the ChatGPT app. Separately, Microsoft is reportedly building its own AI "super app" combining GitHub Copilot, the Copilot chatbot, Copilot Cowork, and a new agentic workflow feature internally called Autopilot. It's a Microsoft-flavored version of what OpenAI is doing with the ChatGPT-Codex-Atlas ecosystem.

And a quick preview: Apple's WWDC kicks off next week, and Greg Joswiak is teasing a very glowy logo that matches rumors of a massive Siri AI upgrade. The expected iOS twenty seven Siri interface is described as a major visual and functional overhaul. With Computex this week and WWDC next week, June is absolutely stacked.

That's the briefing for Tuesday, June second. I'll be back tomorrow. For the Local AI and Automation Edition, stay tuned for episode two in the feed.