Three former OpenAI researchers, Misha L. Balesny, Tomek Korbak, and Jasmine Wang, who were abruptly terminated last week, have officially released a public letter today. In the letter, they openly expressed concerns that the company's sudden action would make other current employees hesitant to speak out, thereby causing serious negative impacts on OpenAI's overall corporate culture.

The public letter, titled "OpenAI Cannot Alone Ensure AI Safety," was posted on social media by Balesny, who previously worked on AI alignment research. The letter is addressed to OpenAI's various safety committees. The other two signatories, Korbak and Jasmine Wang, also previously focused on AI safety research and were terminated along with Balesny last week.

Regarding the specific reasons for the termination, Balesny revealed in his post that the company cited "excessive communication with third-party security agencies" as the cause of the breach of trust. He clarified that he had never leaked any company intellectual property information and had always maintained high coordination with his direct supervisor, the research department management, and the board of directors. Another signatory, Korbak, also stated that the company had verbally informed him of the reason for his termination, citing "the way he communicated with METR." However, he emphasized that communicating with METR was part of his responsibilities.

In response to this incident, the three individuals expressed deep concerns in the public letter, stating that this sudden and hasty termination has already affected OpenAI's previously celebrated open culture, potentially making other employees hesitant to express their opinions or work freely in the future. To address this, the three proposed three specific suggestions: first, allowing third-party security auditors to enter the company; second, ensuring that cutting-edge AI models continue to be effectively monitored; and finally, supporting security researchers in maintaining open and transparent communication with other institutions in the industry.

Facing external scrutiny, a research department official at OpenAI expressed strong support for these three suggestions in an internal memo sent to employees afterward. The official stated that the company greatly appreciated the contributions of the three individuals to AI safety work and their willingness to speak out. They clarified that the termination decision was unrelated to the issues they raised about safety or their public statements, emphasizing that OpenAI always encourages and will continue to encourage employees to express their opinions, and would never terminate an employee for expressing concerns.