The White House convened the biggest names in AI this week, and the headline is less about new regulation and more about a rebrand. Trump hosted executives from OpenAI, Anthropic, Meta, Google, and Nvidia, where they agreed to voluntary safety standards including independent audits. The administration is now pushing the term "Super Intelligence" to describe the technology, a shift that signals a focus on capability over caution. The voluntary framework is a far cry from the binding rules some European regulators are pursuing, but it keeps the industry's biggest players at the table without forcing compliance. Meanwhile, a massive security lapse at Microsoft is a stark reminder of what happens when trust is misplaced. A researcher discovered they could have accessed an estimated 17 trillion records across Microsoft's internal analytics services simply because a login token signature was never verified. The flaw, detailed in a new report, allowed the researcher to claim any identity within the system. It is a textbook example of a single missing check creating a catastrophic vulnerability, and it underscores why the voluntary audits agreed upon at the White House need to be more than just a press release. In other security news, Dutch police arrested a 24-year-old convicted hacker in connection with the ShinyHunters group, a prolific data theft and extortion ring. The arrest, covered by KrebsOnSecurity, suggests law enforcement is making progress in dismantling a network responsible for some of the most high-profile breaches of the last few years. On a lighter but no less technical note, a developer has released a fascinating deep dive into testing WebGPU data layouts with a tool called Facet, offering a practical look at how to share configuration data between host and shader code without the usual headaches.