OpenAI flips on its own bill — now wants stronger AI safety rules

OpenAI has publicly asked California to beef up SB 53, a bill the company spent months fighting against. The shift is stark: what was once an open letter warning the state not to pass it is now a letter begging it to pass something bigger and stricter.

The company's argument is pragmatic. It says the original bill is too narrow, leaving a hole that lets other models slip past the same safety requirements. If only OpenAI has to comply, competitors with less money or different architectures can undercut it — and the whole point of the law evaporates. So OpenAI is asking the state to widen the scope, raise the bar, and close the loophole.

This is one of those rare moments where Silicon Valley's biggest player and the regulators it once bickered with are actually singing from the same sheet. But don't mistake alignment for sincerity. The bill still lands on the companies that can afford compliance teams and safety labs — the ones with real capital behind them. Smaller players and open-source groups will feel the weight of it. And the bill's language will decide whether it's a safety floor or a moat.

Why this matters for us: the companies writing these rules get to set the standard everyone else has to meet, so the ones we rely on — the shops, the crews, the side hustles — end up paying the price whether they asked for it or not.