Jan Leike, a key OpenAI researcher who resigned earlier this month citing “safety concerns,” has joined competitor Anthropic to “work on scalable oversight, weak-to-strong generalization, and automated alignment research,” according to Leike’s post on X.
Anthropic has positioned itself to be the antithesis of OpenAI: a public benefit corporation dedicated to cautious progress. It was founded by ex-OpenAI employees who departed, according to the Financial Times, because they believed OpenAI was straying from its mission of developing safer, more ethical AI and becoming too focused on commercial interests after the $1 billion Microsoft deal. For those following the AI drama, Leike joining Anthropic comes as no surprise.
“We are long overdue in getting incredibly serious about the implications of AGI. We must prioritize preparing for them as best we can,” Leike said in follow-up posts about his resignation a few weeks ago. “Only then can we ensure AGI benefits all of humanity.”
Leike previously worked on OpenAI’s “Superalignment” team, which seems to align with his role at Anthropic, ensuring that intelligent systems are designed to act in accordance with human values. In his thread after announcing his resignation, Leike criticized OpenAI for not providing enough compute resources to ensure success.
For its part, Anthropic employs “constitutional AI,” which incorporates ethical principles into an AI model’s training by gathering public input and integrating concepts from sources such as the United Nations Universal Declaration of Human Rights, as I’ve previously reported.
Another notable departure from OpenAI is its co-founder, Ilya Sutskever, who left earlier this month and had reportedly been absent from the office since he led a failed coup against CEO Sam Altman in November. Although Sutskever resigned and hasn’t yet indicated his next move, his values align with Anthropic’s, making it a plausible destination for him, alongside Leike.