OpenAI has formally acknowledged the external testers who contributed to the safety evaluation of its new 'o1' model series, including o1-preview and o1-mini. These independent experts, spanning AI safety, cybersecurity, and alignment research, provided critical feedback before the model's public release. The acknowledgements appear in the official OpenAI o1 System Card, a document detailing safety protocols, testing methodologies, and risk mitigation strategies.

By involving third-party testers, OpenAI aims to enhance transparency and rigor in its development process, addressing concerns about the potential misuse of advanced reasoning models. This move is part of a broader industry trend where AI developers collaborate with external stakeholders to identify and mitigate risks like chemical, biological, and radiological (CBRN) threats, as well as cybersecurity vulnerabilities. The external testers' insights help shape the model's safety features, such as refusal of harmful requests and adherence to secure coding practices.

This collaborative approach is crucial for building trust and ensuring that powerful AI systems are deployed responsibly, balancing innovation with public safety and ethical considerations in the rapidly evolving AI landscape.