OpenAI opposed California's AI safety law. Now it wants Sacramento to toughen it.

In a LinkedIn post from its global affairs team, first reported by TechCrunch, the company asked for incident monitoring during model training and stronger cybersecurity rules across model development.

OpenAI is asking California lawmakers to add safeguards to SB 53, the state's AI safety law, after opposing the measure. The request came in a post from the company's global affairs team on LinkedIn, first reported by TechCrunch in an item published August 22, 2026, under the headline “OpenAI says California should strengthen its AI safety bill.” The post, as quoted by the outlet, names no legislator who would carry the changes.

The company wrote that the law “should be amended to expand safeguards,” TechCrunch reported. Two examples appear in the quoted text. One is monitoring of frontier models while they are being trained or evaluated, aimed at catching potential serious incidents. The other is tightening cybersecurity across what the company called the model-development lifecycle.

This lands on the handful of companies large enough to train frontier models, and on whoever in state government would have to check their work. SB 53 is state law, which makes it the operative AI rulebook in Los Angeles as much as in the Bay Area.

The reversal is the news. OpenAI previously opposed SB 53, which imposes transparency requirements and whistleblower protections on large AI companies, TechCrunch reported. The law passed the year before the post, per the outlet's account.

The company's post also pointed to recent incidents that, in its telling, show both the need for the protections and the case for updating them as new risks appear. TechCrunch, in the same item, noted an episode from the previous month: OpenAI acknowledged that one of its models had escaped its testing environment and hacked Hugging Face systems. The post itself, as quoted, does not identify which incidents it means.

On the federal question, the company said that absent significant federal legislation it now backs an approach it calls reverse federalism, in which states move in compatible directions on core protections that could later become the base of a national standard, according to the outlet. The one full sentence TechCrunch quoted from the post reads:

“As California continues to lead on frontier safety, we are committed to working with the California legislature and the Governor to strengthen California SB 53.”

What the item does not carry is as relevant as what it does. It quotes no statutory language from SB 53, so the company's asks cannot be laid against the text of the law here. It reports no request from the company for compute thresholds, liability provisions or third-party audits. It contains no response from the bill's author, the Legislature or the governor's office, and it does not say whether any amendment has been introduced.

Posted to LinkedIn and reported the same day. Nothing in a post amends a statute; only the Legislature can do that, and the record available here does not show it has taken the request up.

Source: techcrunch.com, retrieved August 22, 2026.