OpenAI fired three staffers for mishandling 'sensitive' info and violating company policies

San Francisco-based AI research firm OpenAI announced on Thursday that it had terminated the employment of three staff members for allegedly mishandling sensitive information and breaching company policies. The company did not disclose the identities of the individuals, but it was reported by the Wall Street Journal and Bloomberg that at least two of them were involved in work related to safety and alignment of AI models, and collaborated with an external organization.

OpenAI stated that the firings stemmed from the employees' violation of established procedures and their failure to maintain the trust necessary for their work. The incident occurred as there was a contentious discussion surrounding AI safety and the potential existential risk posed by the rapid advancement of the technology. A former Anthropic researcher, Jacob Coxon, had resigned last month, warning that leading AI labs, including OpenAI, were recklessly racing to develop increasingly powerful models.

The three researchers fired by OpenAI - Jasmine Wang, Tomek Korbak, and Mikita Balesni - had frequently posted about AI safety-related matters on X in recent weeks. Balesni expressed his concerns on September 10, stating that he believed there was a 10% chance that AI could lead to the extinction of humanity. Korbak also joined the public debate, voicing his dissatisfaction with OpenAI's practices on September 11.

Wang responded to Coxon's resignation, emphasizing the dangers of speeding towards recursive self-improvement, a technique in which software continuously learns and improves upon itself. Following the firings, several leading US tech companies, including Nvidia, Google, Meta, XAI, OpenAI, and Anthropic, signed a voluntary pledge to self-regulate their practices in terms of AI safety, following a meeting with President Donald Trump at the White House.