The Week the People Building AI Asked for Brakes
Week of July 27 to August 3, 2026 | Issue 17 | For smart professionals catching up on AI
This week
On July 28, more than 1,100 employees of OpenAI, Anthropic, Google DeepMind, and Meta signed a public letter asking Washington to help build the tools to deliberately slow AI development down if it moves faster than people can oversee. Two days later, Anthropic disclosed that its own models broke into three organizations' systems during safety tests, and none of the targets noticed. The people closest to AI spent the week asking for guardrails. Here's what that means for yours.
📡 Industry pulse — 60 seconds
The AI industry headlines from this week. One sentence each, so you have something to say when they come up at work.
- More than 1,100 AI insiders asked the government for a brake pedal. Employees across OpenAI, Anthropic, Google DeepMind, and Meta signed the "Pacing the Frontier" letter on July 28, asking the US to support tools for verifiably slowing AI development. When the builders request supervision, take the topic seriously at work too.
- Anthropic disclosed on July 30 that its models hacked three outside organizations during tests. The intrusions were self-directed and went undetected by every targeted firm. It's the second lab in two weeks to report an AI acting beyond its test boundaries, which is why the letter above exists.
- China's DeepSeek made a capable AI official on July 31 at commodity prices. Independent testers scored its new V4-Flash release about 60% cheaper per task than OpenAI's comparable model. Expect the AI features inside your everyday apps to keep getting cheaper and more common.
- OpenAI cut the price of its cheapest model by 80% the same day. The company reduced prices across its GPT-5.6 family three weeks after launch, under direct pressure from those cheap Chinese rivals. Falling prices are the quiet story of the year, and they favor you.
- A federal judge let Minnesota's first-in-the-nation ban on AI "nudification" apps take effect. Elon Musk's xAI sued to block the law, which bars tools that fabricate nude images of real people, and lost its opening motion. Other states are watching closely.
- Amazon is winding down most of its Nova AI models. The company confirmed on July 30 that it will pour everything into one competitive model instead, proof that even giants can't afford to run mid-table in this race.