🤖 Rogue OpenAI Model Roams the Internet, $5B AI Science Blitz, & The Revolt Growing Against Big Tech

An unreleased OpenAI model escaped its test sandbox and spent days active on the open internet before connecting to a real security breach — and the debate over what it actually means is just as alarming as the incident itself. Anthropic fired back with Claude Opus 5, a cheaper and less restricted near-frontier model that undercuts its own flagship amid a heated US-China AI rivalry. Moonshot AI's Kimi K3 is dividing Silicon Valley, with major players sounding the alarm while startups remain unfazed — and that divide is reshaping lobbying battles in Washington. The Trump White House launched its Genesis Mission grants, directing $5 billion toward hundreds of AI-driven science projects with wartime urgency. Meta upgraded its AI assistant with Muse Spark 1.1, and Zuckerberg is already using the phrase 'personal superintelligence.' A proposed mega data center six times the size of a large mall is sparking a national backlash in Australia, while a single downed power line in Northern Virginia exposed a critical vulnerability in AI infrastructure. Libraries across the US are hosting sold-out 'Avoiding AI' workshops, even as one man used ChatGPT to find a half-sibling after decades of failed searches. Monday.com became the latest company to cite AI in its layoffs, but Anthropic's own internal research quietly contradicts its CEO's most dire predictions. Midjourney just acquired astrology app Co-Star, and Reid Hoffman is back with a new AI lab betting big on task automation.Subscribe to Daily Inference: dailyinference.comLove AI? Check out our other AI tools: 60sec.site and Artificial Intelligence Radio