The Week AI Came Back Online, and the Rules Caught Up
Week of June 29 to July 6, 2026 | Issue 13 | For smart professionals catching up on AI
This week
Three weeks after the US government switched off Anthropic's most powerful AI, it switched it back on. Claude Fable 5 returned worldwide on July 1, and the deal that brought it back is the closest thing yet to a rulebook for frontier AI. The same week, Colorado's AI law became the first comprehensive US state AI regulation to take effect, OpenAI's newest model stayed locked behind government approval, and OpenAI teased its first-ever piece of hardware. Away from the power struggle, the UK started plugging AI into its health hotline, a Stanford study put hard numbers on AI and young workers, and Meta launched an app that builds video games from a sentence. The theme of the week: the rules are catching up to the capability. Here's what that means for you.
📡 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.
- Anthropic's Claude Fable 5 returned worldwide on July 1 after a three-week government shutdown, brought back by a new safety system and an industry jailbreak-scoring framework Anthropic built with Amazon, Microsoft, and Google.
- Colorado's AI Act became enforceable on June 30, the first comprehensive US state law governing AI, focused on transparency and consumer rights rather than heavy compliance paperwork.
- OpenAI's GPT-5.6 stayed locked to government-approved customers, with the White House signing off on access company by company, an arrangement OpenAI publicly called a short-term step, not the future.
- OpenAI teased its first hardware product on June 29, and it's a surprise: a small keyboard "macro pad" for its Codex coding tool, not the consumer AI gadget many expected, launching July 15.
- OpenAI is reportedly weighing an IPO delay to 2027 and, separately, discussed giving the US government a roughly 5% stake in the company.