Big AI Had a Point When It Said It Needed to Be Told What Is Not Okay

So Big AI has finally found something to agree on: they'd like the Trump administration to tell them what's not okay. How quaint. Just weeks after begrudgingly slowing down GPT-5.6 at the administration's 'request,' we now have Dario Amodei telling us he's 'glad' the administration is taking AI seriously — while quietly noting the process is 'abrupt' and 'opaque.' That's code for 'we have no idea what the rules are and it's making investors twitchy.'

The article's title practically writes itself: Big AI had a point. Of course they did. When the only signals from the White House come through backchannel requests and carefully worded 'suggestions', the message isn't clarity — it's a power game. The administration wants the ability to lean on companies without the accountability of explicit legislation. And the labs, having spent years telling regulators to back off, now find themselves begging for a rulebook they can't read.

All this while China's Zhipu is cheerfully open-sourcing models that match Opus at a fifth of the cost, as we noted last week. But sure, let's have a 'thoughtful dialogue' about boundaries while the rest of the world just builds. The current situation is the worst of both worlds: enough regulation to slow down American companies, not enough to actually govern what they build.