Meta Fires AI Safety Experts Just Three Months After Hiring Them Despite Pledging Enhanced Safety

Meta has reportedly let go of specialized personnel it brought in just about three months ago to strengthen the safety of its artificial intelligence (AI) agents.

US online media outlet Semafor reported on the 2nd (local time) that Meta recently laid off employees it had recruited from the AI safety startup Virtue AI.

Andy Stone, a spokesperson for Meta, stated, "Regrettably, this arrangement did not work out as planned," pointing to a clash in working styles as the cause.

However, Stone explained that the Meta Superintelligence Lab (MSL) remains dedicated to AI safety, alignment (keeping AI goals consistent with human values), and risk management.

Previously, Meta recruited Virtue AI co-founders Bo Li, Dawn Song, and Sanmi Koyejo, along with other staff members, in late June.

At the time, online media outlet Axios reported that Meta stated in an internal memo, "As we deliver AI products to billions of users and develop increasingly capable agents, keeping systems stable and reliable is a fundamental requirement."

Virtue AI has previously collaborated with Anthropic and OpenAI, and has a track record of performing AI security tasks with the National Institute of Standards and Technology (NIST) under the US Department of Commerce.

Notably, co-founder Dawn Song is a world-renowned authority in the field of AI safety who serves as a professor at the University of California, Berkeley (UC Berkeley) and co-director of the Center for Responsible Decentralized Intelligence (RDI).

She recently published a joint paper with research executives from major AI companies and academic scholars, warning that the "intelligence explosion" triggered by AI could lead to the extinction of humanity.

Meta's move to lay off AI safety personnel contrasts with companies like Anthropic and OpenAI, which have voiced concerns over AI risks and advocated for the establishment of self-regulatory bodies to slow down development speeds.

Meta CEO Mark Zuckerberg has previously expressed opposition to arguments for slowing down development, stating that development companies themselves can set appropriate paces without external regulation.

He is also known to have conveyed his opposition to the establishment of a self-regulatory body to President Donald Trump, playing a part in scuttling the plan.
※ Please note: This article was translated by AI and may contain errors.