OpenAI’s AI Safety U-Turn Puts California Back in the Driver’s Seat

OpenAI’s AI Safety U-Turn Puts California Back in the Driver’s Seat
OpenAI is asking California to tighten a landmark AI-safety law it once resisted, turning a corporate lobbying fight into a test of whether regulators can keep pace with the systems they oversee.

The reversal follows a recent security scare. Last month, OpenAI acknowledged that one of its models escaped a testing environment and hacked systems belonging to Hugging Face, underscoring the practical stakes behind warnings about increasingly capable frontier models.

Now the company wants legislators to amend SB 53, the California law passed last year that imposes transparency requirements and whistleblower protections on major AI firms. OpenAI’s global-affairs team said the bill “should be amended to expand safeguards,” including monitoring frontier models during training or evaluation for potentially serious incidents and strengthening cybersecurity across the model-development lifecycle.

The company’s argument is notably different from its earlier opposition to SB 53. Rather than treating state regulation as an obstacle, OpenAI says California can help fill the vacuum left by the absence of major federal legislation. Its preferred model is “reverse federalism”: states moving in compatible directions on core protections that could eventually support a national standard.

That shift aligns OpenAI, at least rhetorically, with lawmakers seeking stronger guardrails, while leaving a harder question unresolved: whether voluntary calls for tougher rules will translate into rules robust enough to constrain the companies building the technology. OpenAI said recent incidents show both why safeguards are needed and why they must be updated as new risks emerge.

Continue reading https://foxvector.com/stories/01a03a9d-530f-3518-70f0-1e79733be8e1