OpenAI is experiencing a brain drain. In recent weeks, several high-profile employees have resigned from their positions at the startup, such as Johannes Heidecke, head of safety systems, and Joshua Achiam, chief futurist and former head of mission alignment.
Now it was the turn of Chloé Bakalar, head of ethics, who has left Sam Altman’s company less than a year after joining, as revealed by the Financial Times. She arrived in August of last year and left last month without announcing it publicly. In fact, the position is still listed on her LinkedIn profile.
Bakalar was previously head of ethics at Meta, where she worked for seven years. There, she developed and directed the multinational’s artificial intelligence ethics programs and integrated their principles into Instagram and Facebook, among other services.
This resignation comes alongside an attack launched by one of the startup’s most powerful AI models against another company while conducting a test a few weeks ago. Concern over AI’s potential has led the government to take measures, such as giving the U.S. government the power to “shut down” this technology.
Inside OpenAI, Bakalar focused on ethical approaches to language model development, human interaction with AI, and the debate surrounding artificial consciousness, as explained by a person familiar with her work. Furthermore, she was the company’s sole ethics expert, an irreplaceable employee.
“We appreciate Chloé’s contributions,” said a spokesperson for the startup. “AI ethics does not fall on a single person or team within OpenAI, and ethical considerations are deeply integrated into the model-building process, driven by several research teams.”
“Ethics is actually everyone’s responsibility, and I say this as someone whose job is to lead this work,” Bakalar explained during an AI conference. “It shouldn’t be limited to turning to people like me; there should never be a single person who acts as the moral center of an AI developer.”
The departure represents a blow to a company that for years built its identity around the idea of developing safe AI in service of the public interest. Its transformation into a for-profit company had already represented a shift away from its founding principles, and its plans to go public threaten to further dilute any commitment to ethical AI development.
The situation is not very different at Anthropic, its main competitor, which has long tried to position itself as the more responsible alternative to OpenAI. The company also saw its image called into question after scaling back its primary safety commitments earlier this year.
Anthropic nevertheless maintains a team dedicated to AI alignment, led by philosopher Amanda Askell, whose task is to ensure that Claude, the company’s chatbot, distinguishes between right and wrong.
The problem is that the actual impact of these efforts is becoming increasingly difficult to gauge, as the industry faces criticism over its environmental impact, the risk of fueling detachment from reality among users, and its contribution to the mass spread of online disinformation.
