Dismissals at OpenAI Raise Questions About Balance Between Transparency and Secrecy in AI Research

Three former safety researchers at OpenAI have disputed their dismissals in an open letter to the company.

The researchers, Jasmine Wang, Mikita Balesni, and Tomek Korbak, claim that the public nature of OpenAI's communications around their firing may be having a chilling effect on employees who are crucial to ensuring AI safety.

Their dismissal came after they allegedly shared information with an external AI safety organization. This move has raised questions about the balance between transparency and secrecy in the field of AI research, particularly when it comes to sensitive company information.

The researchers point out that they were previously encouraged to raise safety concerns and collaborate with independent organizations, which was seen as a key aspect of OpenAI's culture.

However, their open letter suggests that this may no longer be the case, and that employees are now being discouraged from speaking out or sharing their expertise.

The researchers who were fired from OpenAI have claimed that they acted in accordance with the company's mission and internal norms at the time of their dismissal.

They also addressed speculation about the leak of a story involving OpenAI's new architectures, stating that they were not responsible for the leak and did not engage with external parties outside their official roles. They emphasized that any communication with outsiders was done in coordination with senior executives and board members.

The former employees expressed concerns that the firing may be indicative of shifting norms within the company, leaving employees uncertain about their role and responsibilities. This shift could potentially undermine OpenAI's commitment to embedding third-party safety auditors.

In a final statement, the researchers recommended that OpenAI adhere to its public promises regarding transparency and collaboration with external safety experts, and maintain the monitorability of its frontier models.

OpenAI's leadership has seemingly acknowledged the concerns raised by the dismissed safety researchers, but expressed reservations about the company's openness.

The researchers' open letter has sparked a public discussion about OpenAI's handling of employee departures and its relationship with external safety experts. Wang warned that unless action is taken to address these issues, she fears that other employees may face similar treatment in the future without clear explanations.