Loading page…
Meta attributes the split to working-style differences and says its AI lab remains focused on safety. The recruits brought experience in testing and safeguarding AI systems.
Listen to this story
Meta has ended its employment arrangement with the Virtue AI specialists it recruited, including the startup’s three co-founders, roughly four months after the June 2026 hires. Spokesperson Andy Stone blamed differences in working styles, but the nature of those differences and the terms of the arrangement remain undisclosed. The split reverses a move intended to add AI security expertise; Meta says its Superintelligence Labs still prioritizes safety, alignment and frontier risk.
Semafor reported the split on October 2; Virtue AI had not commented when the report was published.
The hires included co-founders Bo Li, Dawn Song and Sanmi Koyejo, along with other Virtue AI employees.
Virtue AI’s enterprise tools included automated red teaming, runtime guardrails and AI governance.
Meta is letting go AI security specialists just four months after bringing them into the company. The employees joined from Virtue AI in June 2026, according to Semafor’s October 2 report. The decision removes recruits from a startup whose work included testing AI systems for weaknesses and developing safeguards.
Meta spokesperson Andy Stone attributed the separation to clashing working styles. What those differences involved remains unclear. The Virtue team did not immediately respond to Semafor’s requests for comment when it published the story, leaving the employees’ perspective on the split unknown.
Unfortunately, the arrangement didn’t work out as planned.
Andy Stone, Meta spokesperson, speaking to Semafor
Stone said Meta Superintelligence Labs remains focused on AI safety, alignment and frontier risk. His statement maintains the lab’s commitment to those areas even as Meta ends its employment arrangement with the specialists it recruited.
The original hiring included Virtue AI co-founders Bo Li, Dawn Song and Sanmi Koyejo, alongside other employees. Axios reported their move in June, according to Semafor. The terms of the hiring arrangement were not disclosed, The Times of India reported.
The Times of India described Virtue’s enterprise security tools as covering automated red teaming—tests designed to identify weaknesses in AI systems—as well as runtime guardrails and AI governance. The startup’s work focused on identifying potential risks and developing tools to test and safeguard systems.
The hiring was intended to strengthen Meta’s AI security efforts, including security around AI agents, according to the newspaper’s account of Axios’s reporting. It quoted a hiring memo describing safe, reliable and trustworthy systems as foundational as Meta built increasingly capable agents.
Virtue had previously worked with Anthropic, OpenAI and the Commerce Department’s National Institute of Standards and Technology, known as NIST. Its experience therefore spanned work with AI developers and a federal standards agency.
Loading discussion...
Join the conversation
Explain how you would balance public accountability with employee privacy.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.