Meta hires Oasis founder Dawn Song for AI safety push

UC Berkeley professor and Oasis Labs founder Dawn Song has joined Meta Superintelligence Labs as vice president of AI research. 

She said she will help lead Meta’s AI safety and AI security efforts. Song announced the move in a post on X. She said several members of the Virtue AI team will also join Meta. Axios also reported that Virtue AI co-founders Bo Li and Sanmi Koyejo are among the hires.

Song said her work at Meta will focus on frontier AI models and agentic AI systems. She wrote that AI must be “secure, trustworthy, and beneficial” if it is to reach its full use.

The move gives Meta more senior talent in AI security. It also brings a well-known blockchain privacy researcher into one of the world’s largest AI labs.

Virtue AI team moves to MSL

Virtue AI was founded in 2024 to build tools for trustworthy AI. Song said the team worked on AI security, agent security, benchmarks and open platforms before the Meta move.

According to Axios, Meta is hiring several Virtue AI leaders and team members. The report said the group worked on automated red teaming, runtime guardrails and AI governance.

Meta’s interest comes as AI labs put more attention on agent safety. AI agents can take actions, use tools and handle tasks across software systems. That makes security more important because errors or misuse can spread across real products.

As previously reported, Meta has been building a superintelligence AI team after its large Scale AI deal. The company wants to improve its AI models and ship them across Facebook, Instagram, WhatsApp and other products.

Oasis background adds crypto angle

Song is also known in crypto as the founder of Oasis Labs. The company raised $45m in 2018 to build privacy-first cloud computing on blockchain. Its backers included a16zcrypto, Accel and Binance Labs.

The Oasis project later became tied to the Oasis Network and the ROSE token. The network focuses on confidential computing, data privacy and privacy-preserving applications.

In a previous article, crypto.news discussed Oasis Protocol’s verifiable AI agents for crypto trading. The project used trusted execution environments to keep strategies private while giving users proof of how agents behave.

Previously, crypto.news explored Oasis-based AI and data services through Pontus-X, a platform built around privacy and data control. Song’s Meta role connects that same privacy and security theme to a much larger AI platform.

ROSE remains near record lows

The hiring news has not changed ROSE’s weak market setup. Oasis traded near $0.0059 on June 26, close to its intraday low. That is about 99% below its all-time high near $0.596.

ROSE has also struggled with the broader crypto market selloff. Its market value remains far below peak-cycle levels, even as AI and privacy remain active themes in the sector.

The move is still notable for the Oasis community because Song helped shape the project’s early research identity. Her work linked blockchain, privacy and security before AI safety became a major mainstream topic.

For Meta, the hire adds academic and startup experience to its AI safety push. For crypto, it shows how privacy and security talent from blockchain continues to move into frontier AI.