The AI policy battle is getting messy

States won’t wait for federal AI rules

California just moved ahead on AI rules despite direct federal pressure.

Gov. Gavin Newsom issued an order requiring safety and privacy standards for AI companies working with the state, even as the White House warned states to stop regulating. More than 100 state laws are already in place, covering things like child safety and AI system testing.

This is turning into a real power struggle, not a policy debate. The federal government wants a single national approach, but states see a gap and are filling it fast.

The result is friction that companies can’t ignore, especially those selling into public sector contracts. You can already see the split forming between speed and oversight, and neither side looks ready to back off.

It’s mid-morning in Sacramento, and lawmakers are drafting rules while federal officials talk about blocking them.

If this keeps going, companies may end up building for 50 different rulebooks anyway.

Game studios hand content production to AI

Most game ideas fail, and Sett is building around that.

Sett just raised $30 million to expand a platform that generates ads and in-game content using agent-based AI systems. It plugs into studio data and continuously tests variations across marketing creatives and gameplay features.

This feels less like a creative tool and more like a volume machine tuned by data feedback loops. If 97% of ideas fail, then the obvious move is to produce far more ideas and let performance decide.

That shifts power away from creative instinct and toward systems that can run thousands of tests without slowing down. You can hear the logic in it, especially in mobile where ad performance already dictates what gets built next.

I think this pushes games toward safer, optimized content that looks different on the surface but behaves the same underneath.

At what point do players start to notice the pattern?

Mistral is building its own AI backbone

Mistral is borrowing big to control its own infrastructure.

The company raised $830 million in debt to build a data center near Paris, expected to go live in Q2 2026, powered by Nvidia chips. It’s part of a broader plan to reach 200 megawatts of compute across Europe by 2027.

This is a clear move away from relying on cloud providers and toward owning the full stack, even if it means taking on heavy financing. Debt at this scale suggests confidence in long-term demand, not quick returns.

Europe has been talking about AI independence for a while, and this is one of the more concrete steps so far. If Mistral pulls this off, it gives governments and enterprises a local option that isn’t tied to U.S. infrastructure.

The risk is obvious in the hum of a data center floor at night, where costs run nonstop whether usage shows up or not.

Will enough customers choose control over convenience?

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.