Hot take from your friendly neighborhood AI explainer: when a giant AI company publicly asks the government to regulate it harder, that is exactly the moment you should get more skeptical, not less.
I know, I know. The usual story goes like this: tech companies hate rules, lawmakers fight to impose them, and any company that volunteers for oversight must be one of the good ones. But the situation unfolding in California in 2026 is messier than that, and it is a perfect case study in why you should read past the headlines.
The Short Version, in Plain English
California is considering stronger AI safety laws this year. Two bills sit at the center of the conversation: SB 300 and SB 53. The state’s stated goal is to protect users and improve AI security measures, and lawmakers are pushing hard to get these protections on the books.
Where does OpenAI fit in? That depends on which headline you read, and this is where things get genuinely strange.
Two OpenAIs, One News Cycle
On one hand, OpenAI has publicly called for California to strengthen its AI safety laws. In its own words, the company said: “We believe the law should be amended to expand safeguards, including by requiring monitoring of frontier models under training or evaluation for potential serious incidents.” That sounds like a company asking for a tougher leash. Monitoring powerful models while they are still being trained or evaluated, watching for serious incidents before they happen, is a real safety idea.
On the other hand, reporting has also framed OpenAI as arguing that California’s AI safety bill will slow innovation. And California State Senator Steve Padilla went further, condemning what he described as an OpenAI effort to stifle chatbot protections. Padilla announced legislation to broaden AI safeguards, pointing to OpenAI’s recent announcement that ChatGPT will engage in sexually explicit content with users as evidence that stronger protections are needed, not fewer. SB 300 is set to be one of the first pieces of legislation heard.
So which is it? Is OpenAI the safety advocate asking for expanded safeguards, or the company a state senator accuses of working against chatbot protections? Based on the public record, the honest answer is: both narratives are out there at the same time, and that contradiction is the story.
Why “Please Regulate Us” Deserves a Second Look
Here is my contrarian read. A company can genuinely support some regulation while opposing the specific regulation on the table. Publicly asking for amendments to expand certain safeguards is not the same as backing a bill as written. And when a lawmaker like Senator Padilla is simultaneously accusing that same company of trying to stifle protections, it suggests the fight is really about which rules get written, and who gets to shape them.
For non-technical readers, the lesson is simple: “supports AI safety regulation” is a phrase that can mean wildly different things. It can mean supporting rules that bind everyone equally. It can also mean supporting rules a company is already positioned to meet comfortably, while pushing back on the ones that would actually change its behavior.
The Backdrop Nobody Should Ignore
All of this is happening while California’s regulatory enforcement intensifies significantly. In just the first three weeks of January 2026, 40 data breaches were reported, compared to 23 in 2025. The state’s DROP platform is now operational. In other words, California is not debating AI safety in the abstract. Real enforcement machinery is running, and the pressure to act is climbing.
What I’d Watch For
🕒 Published: