pnut

look at this shit….

Greg Brockman says OpenAI pointed Astra at its own systems until it ran out of vulnerabilities to find: "We took 25% of our production engineers and said, 'Sorry, all your projects are on hold. You are now defending. You are now up-leveling our security architecture. You're going to use the models to find all the holes.' And we found a number of serious issues, and we fixed them." "We found some new problems, but eventually it saturated. We basically have found, to our knowledge, all of the P0s, all of the critical problems that Astra is smart enough to find. And of course, there will be a new model, there will be a new round." "You want to be in this tight loop of new cyber capability drops, you deploy it against your systems, you find the new holes, and ideally, you've managed to automate this, what we call defense factory. That's what we're building internally." "There are ideas, for example, formally verifying all of software, that are possible with AI." @gdb @bhorowitz

I was part of a group that worked on that problem. Sure, if you order the people going in you can get great results with your scheme. The problem is that all time you save on boarding is lost in trying to organize preboarding. Imagine trying to line up every passenger.

Microsoft Threat Intelligence has observed Storm-2945, a sub-cluster of Midnight Blizzard, compromising hospitality-related networks worldwide to steal credentials, access cloud environments, and deliver malware to travelers. msft.it/6010aBX6s Since early May 2026, the threat actor has conducted widespread but targeted traffic manipulation attacks involving networks served by captive portals. In some cases, users were redirected through actor-controlled phishing infrastructure, while other activity led to the delivery of malware on impacted systems. Microsoft assesses Storm-2945 to be an operational sub-cluster of Midnight Blizzard and details the malware, tradecraft, and cloud-focused techniques used throughout the campaign, including device code phishing and credential theft activity targeting travelers. Get detections, mitigation, and hunting guidance from this Microsoft Threat Intelligence blog post.

In Germany, a talented 14-year-old earns his club money. In America, his parents pay the club $15,000 a year. That single inversion explains why "we will not" is the most accurate line ever written about US soccer. FIFA built a global system for this. Training compensation and solidarity payments send a cut of every transfer fee back to the clubs that developed the player, from age 12 onward. Develop one future pro and your academy gets paid for a decade. Barcelona's La Masia, Ajax, every Bundesliga academy runs on this logic. The kid is the asset. US Soccer refuses to enforce those rules. When Seattle's Crossfire Premier claimed its $60,000 share of DeAndre Yedlin's transfer to Tottenham, it got nothing. Claims on the Dempsey and Bradley transfers died partly because the federation couldn't even produce the youth training records. So American clubs earn zero dollars when a kid turns pro. They earn when a kid enrolls. Which makes the parent the customer, and the product is whatever keeps the parent writing checks: travel tournaments, hotel weekends, $500 showcase events, private training at $100 an hour. Elite pathways run $8,000 to $20,000 a year. A comparable academy spot in Italy costs about 120 euros. Follow the incentive one level deeper and it gets darker. A club dependent on fees can't cut its weakest paying players, so rosters optimize for retention over development. The scouting pool shrinks to families who can afford the cliff, which appears around age 11, exactly when development matters most. The country runs a talent filter sorted by household income instead of ability. Every four years someone proposes fixing this. The proposal always requires the people profiting from the $15,000 model to vote themselves out of business. They will not.