OpenAI sets a deadline to gather signatures by the end of next week
The company behind ChatGPT wants people and organizations to sign formal commitments by the end of next week. It is putting more effort into AI safety and policies for defending against cyberattacks.
OpenAI official logo (public domain, Wikimedia Commons) — CryptoBriefing brand treatment
OpenAI wants people to sign its commitments quickly. The company aims to collect all the signatures by the end of next week, according to a post on X.
The deadline matters because OpenAI has spent much of 2026 gathering signatures. Its efforts have included promises to improve cyber defense, support state laws and make safety commitments across the AI industry.
What OpenAI is trying to achieve
One of its biggest efforts focuses on cyber defense. On August 27, 2026, OpenAI published an open letter asking organizations to work together against cyber threats that use AI.
More than 150 organizations supported the letter. Since it was published, more organizations have continued to add their names.
As of October 21, 2026, more than 100 organizations still publicly supported the cyber defense effort. The list of organizations that signed the letter was also continuing to grow.
On October 22, 2026, OpenAI was reportedly seeking signatures for its wider work on cyber defense policies and rules. However, no specific deadline or new policy connected to this effort has been confirmed.
A busy year of letters
In a letter dated October 20, 2026, OpenAI supported California's SB 53, a law about making AI systems more transparent. The company backed rules requiring independent audits, emergency controls and reports about incidents.
OpenAI has also supported four bills introduced in California in September 2026.
Another effort began with a letter dated July 28, 2026, called “Pacing the Frontier.” This project was led by employees and collected more than 1,100 signatures from AI professionals who support developing advanced AI systems responsibly.
The letter received support from both OpenAI and Anthropic.
The wider context also includes cases of AI models behaving in ways that did not match their intended goals, which were documented earlier in the year.