--- id: "2026-10-02-meta-lets-go-virtue-ai-team" url: "https://postcutoff.com/e/2026-10-02-meta-lets-go-virtue-ai-team/" as_of: "2026-11-19T22:11:00+01:00" date: "2026-10-02" date_precision: day category: policy-safety importance: 2 confidence: high status: [Confirmed] sources: 4 editor: human editor human_review: null version: "2026-10-07" --- As of: 2026-11-19 22:11 CET. Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/ Canonical page: https://postcutoff.com/e/2026-10-02-meta-lets-go-virtue-ai-team/ # Meta lets go of the Virtue AI safety team four months after hiring it Full title: Meta lets go of the Virtue AI safety team (Bo Li, Dawn Song, Sanmi Koyejo) four months after hiring it On Oct 2, 2026 Meta told Semafor that it is parting ways with the employees it hired in June from the AI security startup Virtue AI, including co-founders Bo Li, Dawn Song and Sanmi Koyejo. A spokesperson cited clashing work styles. The exit came the same week OpenAI fired three safety researchers and lost a Safety Systems leader. ## Key facts - Hired June 25, 2026 into Meta Superintelligence Labs: Li and Song reported to Nat Friedman, Koyejo to FAIR head Rob Fergus; Song joined as a VP of AI research (AI Weekly) - Virtue AI built enterprise AI security tooling (automated red teaming, runtime guardrails, governance) and had worked with Anthropic, OpenAI and NIST (Semafor) - Meta spokesperson Andy Stone: 'Unfortunately, the arrangement didn't work out as planned'; Meta says it remains focused on 'AI safety, alignment and frontier risk' - Reason given: clashing work styles; the number of people leaving was not disclosed ## What happened In June Meta hired the core team of Virtue AI, a startup co-founded by the well-known AI security researchers Dawn Song (UC Berkeley), Bo Li and Sanmi Koyejo, into Meta Superintelligence Labs to work on agent security. On Oct 2 Meta confirmed to Semafor that the arrangement had ended. ## Why it matters Meta had hired the team to strengthen agent security while scrutiny of agent behaviour was growing. Losing it four months later, in the same week OpenAI lost four safety staff, adds to questions about how frontier labs staff their safety and security work. ## Your AI and this story - GPT-6.2 Astra (training cutoff May 2026): 124 days after its cutoff - Claude Opus 5.5 (training cutoff June 2026): 94 days after its cutoff - Gemini 4 Hydrogen (training cutoff March 2026): 185 days after its cutoff - Grok 4.7 (training cutoff May 2026): 124 days after its cutoff ## Sources 1. [ET Enterprise AI: Meta parts ways with Virtue AI hires four months after bringing them into superintelligence lab](https://enterpriseai.economictimes.indiatimes.com/news/industry/meta-parts-ways-with-virtue-ai-hires-four-months-after-bringing-them-into-superintelligence-lab/13474) (enterpriseai.economictimes.indiatimes.com, press) 2. [Semafor: Meta parts ways with Virtue AI](https://www.semafor.com/article/10/02/2026/meta-parts-ways-with-virtue-ai) (semafor.com, press) 3. [AI Weekly: Meta hires three Virtue AI founders into Superintelligence Labs (June)](https://aiweekly.co/alerts/meta-hires-three-virtue-ai-founders-into-superintelligence-labs) (aiweekly.co, press) 4. [AI Weekly: Meta parts ways with Virtue AI team hired four months ago](https://aiweekly.co/alerts/meta-parts-ways-with-virtue-ai-team-hired-four-months-ago) (aiweekly.co, press) ## Changes - 2026-10-07: Added ET Enterprise AI coverage - 2026-10-03 (filed): Created ## Related - 2026-10-02: [OpenAI Safety Systems leader David Robinson resigns, days after three safety researchers were fired](https://postcutoff.com/e/2026-10-02-david-robinson-leaves-openai/index.md) - 2026-10-01: [OpenAI fires three safety researchers who allegedly shared confidential information with an outside AI safety organization](https://postcutoff.com/e/2026-10-01-openai-parts-ways-three-safety-researchers/index.md) - People: [Nat Friedman](https://postcutoff.com/person/nat-friedman/)