Google stopped a zero-day hack that it says was developed with AI
This is the one that should be on the front page of every newspaper: Google’s Threat Intelligence Group caught and killed a zero-day exploit that was built using AI, intended to blow past two-factor authentication in a “mass exploitation event.” We’ve spent years debating whether AI-powered cyberattacks were theoretical — they’re not anymore. The arms race is officially live, and the only reason this didn’t end badly is that defenders were also running AI.
OpenAI just released its answer to Claude Mythos
OpenAI’s new “Daybreak” initiative uses its Codex Security AI agent to map threat models from your actual codebase, find likely attack paths, and automate vulnerability detection. Given that Google just stopped an AI-built exploit in the wild, the timing here is either perfectly calculated or very lucky — either way, the pivot toward AI-native security tooling feels less like a product launch and more like a necessary response to the threat landscape we just entered.
Ilya Sutskever Stands by His Role in Sam Altman’s OpenAI Ouster: ‘I Didn’t Want It to Be Destroyed’
The Musk v. Altman trial just produced its most compelling testimony yet: Ilya Sutskever, the man who arguably fired the starting gun on the most dramatic corporate implosion in tech history, took the stand and defended OpenAI while distancing himself from Musk’s framing. “I didn’t want it to be destroyed” is doing a lot of work as a defense for the guy who temporarily destroyed it — but here we are. Whatever you think of the players, this trial is becoming the definitive oral history of how the most important AI company on earth nearly ate itself.
Here’s what Mira Murati’s AI company is up to
Thinking Machines is building what it calls “interaction models” — AI that processes audio and video input while simultaneously generating a response, so the conversation feels less like a chatbot and more like an actual human exchange. The framing is ambitious: “collaborate with AI the way we naturally collaborate with each other.” Murati has been quiet since leaving OpenAI, so this is the first real signal of what she’s actually building — and it’s a direct shot at the turn-taking limitations that make current voice AI feel awkward and robotic.
I Work in Hollywood. Everyone Who Used to Make TV Is Now Secretly Training AI
A working screenwriter lays it out plainly: in eight months, they completed 20 AI training contracts for five different platforms, describing it as “soul-crushing” work that’s become the new waiting tables for the industry AI disrupted. The bitter irony writes itself — the writers who were replaced by AI are now teaching AI to write better, fueling the very system that displaced them. This is the piece that cuts through every optimistic “AI creates new jobs” talking point.
GM just laid off hundreds of IT workers to hire those with stronger AI skills
General Motors quietly laid off hundreds of IT workers and is replacing them with people focused on AI-native development, prompt engineering, and agent workflows. This is no longer a tech-sector story — when the automaker that makes Silverados is restructuring its workforce around AI agents, the labor displacement conversation has gone fully mainstream. “Prompt engineering” jobs at GM weren’t on anyone’s 2020 bingo card.
CUDA Proves Nvidia Is a Software Company
Wired makes the case that Nvidia’s real moat isn’t the H100 or the Blackwell chip — it’s CUDA, the programming platform that’s so deeply embedded in every AI workflow that switching away from it would mean rewriting years of code. It’s a smart read that reframes the competitive landscape: you can build a better GPU, but you cannot easily build a better decade of developer lock-in. AMD and Intel know this. It’s why they’re losing.
Students Boo Commencement Speaker After She Calls AI the ‘Next Industrial Revolution’
At the University of Central Florida, humanities graduates responded to a commencement speaker’s AI boosterism with audible boos and shouts of “AI SUCKS!” — which is honestly more coherent feedback than most commencement speeches get. You can debate the sentiment, but you cannot deny the symbolism: the generation entering the workforce right now, carrying student debt into an AI-disrupted job market, is not buying the “it’s the next industrial revolution!” pitch. Hard to blame them.
Bottom Line
Whether it’s Google stopping AI-built exploits, GM clearing out IT desks for prompt engineers, or screenwriters secretly training the models that replaced them, today’s news makes one thing uncomfortably clear: the transition is no longer coming — it’s already mid-stride.