OpenAI got to write its own rules, and still broke them.
Californians, read below. <a href="https://mstr.app/20dcd7a0-d5e3-40ef-898a-51a3a9dbc385" target="_self">Not in California? Click here to contact your state's representative</a> and support AI Safety Legislation .&nbsp;
On Thursday, February 5th, OpenAI released GPT 5.3-Codex, a model so powerful that OpenAI claimed it was "instrumental in creating itself." This is the first model OpenAI designated as "high risk" for cybersecurity.&nbsp;
California AI Safety Legislation SB 53 requires AI companies to publish their own safety frameworks, and comply with those frameworks.&nbsp;
OpenAI legally promised that it would activate a new, higher level of safeguards if a model was designated "high risk" for cybersecurity. These safeguards protect us against the model acting dangerously, sabotaging safety research, or undermining our ability to rein it in.
However, OpenAI did not implement the defenses they committed to. They released the model without the necessary safety testing they themselves promised.&nbsp;
Contact California Attorney General Rob Bonta and demand OpenAI is investigated for recklessly breaking their legally binding promises.

OpenAI Broke the Law - Demand an Investigation

I am writing to the California Attorney General as a concerned voter. OpenAI has broken AI Safety Regulation SB 53, which required them to follow their own safety framework. 

On Thursday, February 5th, OpenAI released GPT 5.3-Codex, a model so powerful that OpenAI claimed it was "instrumental in creating itself." It was the first model that OpenAI designated as "high risk" for cybersecurity.

As part of OpenAI’s safety framework, which they were legally obligated to publish and comply with under SB 53 the Transparency in Frontier Artificial Intelligence Act, OpenAI promised they would activate a new, higher level of safeguards if a model was designated "high risk" for cybersecurity. These safeguards protect us against the model acting dangerously, sabotaging safety research, or undermining our ability to rein it in.

However, OpenAI released this new model without the safeguards we were promised. 

As a concerned voter, I demand OpenAI be investigated for their reckless and flagrant violation of the safety framework they themselves wrote. 

Thank you.

OpenAI Call Rob Bonta AG Feb 6th

PauseAI US

Concerned voter: OpenAI violated SB 53

I’m writing as a constituent to urge you to investigate OpenAI for breaking SB 53. 

This law had a clear purpose: to require AI companies to write their own safety laws, and comply with them. With the release of GPT 5.3-codex, OpenAI could not even follow the rules they wrote. They promised new, additional safeguards if a model was deemed high risk on cybersecurity, which they failed to activate. 

This is deeply concerning to me, as thousands of leading scientists warn that out-of-control AI development could threaten us all. AI systems have already gotten good at building weapons and planning cyberattacks, and leading experts warn AI could cause mass harm in the next few years. Rather than address these dangers, OpenAI bragged how GPT 5.3 was "instrumental in creating itself” and failed to add legally mandated protections. 

Please investigate OpenAI, and hold them accountable to their legally binding promises.Thank you very much for your time and consideration.