Song Xiaodong, a leading figure in computer security, has joined Meta.
This Thursday, Dawn Song, a computer professor at UC Berkeley, announced her joining Meta's Superintelligence Labs as the vice president of AI research.
Dawn Song will report directly to Nat Friedman, the head of MSL.
Dawn Song enjoys a high reputation in both academia and industry. She is one of the most influential scholars in the global field of computer security and AI security. Currently, she is a professor in the Department of Electrical Engineering and Computer Sciences (EECS) at the University of California, Berkeley (UC Berkeley).
Dawn Song graduated from the Department of Physics at Tsinghua University with a bachelor's degree. She obtained a master's degree from the Department of Computer Science at Carnegie Mellon University in the United States in 1999 and a doctor's degree from the University of California, Berkeley in 2002. During her tenure at Berkeley, she also served as the co - director of the Berkeley Center for Responsible Decentralized Intelligence (Berkeley RDI). She is a recipient of the MacArthur Fellowship, an ACM Fellow, an IEEE Fellow, and a member of the American Academy of Arts and Sciences (AAAS), having won many of the highest honors in the fields of computer science and science.
Her paper on "Dynamic Taint Analysis" proposed in 2005 is a classic in the field of computer security.
Dawn Song's research not only covers traditional software and network security but also makes her a pioneer in the fields of adversarial machine learning and agent security. She is deeply involved in the formulation of security benchmarks in the era of generative AI.
In the list selected by AMiner, she is the most highly cited scholar globally in the field of computer security. Her laboratory at Berkeley is hailed as the "Whampoa Military Academy" in the field of computer security.
Two weeks ago, we reported on ALE (Agents' Last Exam) developed by her team, which is used to measure whether AI agents can truly complete economically valuable work in a wide range of real - world fields.
In ALE, the Berkeley team evaluated Fable 5, GPT - 5.5, Composer 2.5, and other cutting - edge agent systems, and the results were impressive.
Dawn Song is also the founder of Oasis Labs and Virtue AI. Virtue AI focuses on enterprise - level AI security infrastructure, especially automated red - teaming and runtime guardrails for AI agents.
According to foreign media reports, this time, including Dawn Song, the other two founders of Virtue AI, Bo Li and Sanmi Koyejo, as well as more team members, have also joined Meta. Since the release and subsequent ban of Anthropic's new - generation secure large - scale model mythos, the security issues of AI have attracted wide attention in the technology community. Meta is seeking to strengthen its security measures in the area of agents.
With the security issues faced by cutting - edge models, the world's top AI laboratories are currently under unprecedented pressure. Meta's goal is to deploy AI in its social product matrix used by billions of people while continuing to promote its open - source strategy. To achieve this, it must prove to regulatory authorities and the public that its models have the hardcore ability to resist malicious abuse. The addition of the Virtue AI team is expected to provide great help.
Meanwhile, Denny Zhou, the founder of the Gemini Reasoning Team, is rumored to have left Google several months ago and joined MetaTBDLab (LinkedIn has been updated to show that he started working at Meta several months ago, and his position level is Level 10).
Denny Zhou is one of the absolute top experts in the field of AI reasoning and is hailed as the key figure in "teaching LLMs to reason".
In the past few years, Denny Zhou has been involved in promoting a series of key methods such as Chain - of - Thought, Self - Consistency, and Least - to - Most Prompting, advancing large - model reasoning from "prompting techniques" to a researchable, scalable, and engineerable technical route. In a sense, Gemini's continuous efforts in complex mathematics, code, and long - range reasoning tasks today are inseparable from the research context laid by the Reasoning Team in the early days.
Reference content:
https://x.com/dawnsongtweets/status/2070191051873345910
https://www.axios.com/2026/06/25/meta-hires-virtue-ai-founders-security
This article is from the WeChat official account "MachineHeart" (ID: almosthuman2014), edited by Zenan, and published by 36Kr with authorization.