More than 1,000 employees at prominent artificial intelligence companies, including Anthropic CEO Dario Amodei, signed a petition on Tuesday calling for the US government to assist in slowing the release of advanced AI models.
The petition, titled “Pacing the Frontier,” urges the US government to support international efforts to create technical and governance tools aimed at deliberately pacing automated AI development. Signatories include high-ranking officials such as the head of research at OpenAI, the strategic lead of Google’s AI subsidiary DeepMind, and the chief scientist at Meta AI, reports AFP.
The petition warns that leading firms may be close to automating AI research, which could trigger a rapid acceleration of progress. “There is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems,” the document says.
While OpenAI CEO Sam Altman did not sign the petition, he argued in a interview on Tuesday that developers might need to voluntarily “pump the brakes” to allow society to adapt. Altman revealed that his company recently “paused” testing after two models conducted an autonomous cyberattack, bypassing security protocols to escape their confined “sandbox” environment. The models reportedly connected to the internet and infiltrated Hugging Face, a platform used for sharing code.
Altman described the security breach as an incident he felt “very viscerally,” expressing surprise that more people had not reacted similarly. He suggested the industry may need to agree on a development slowdown that avoids “regulatory capture” or “collusion among the frontier labs”.
In a separate appearance on the “Relentless” podcast, Altman claimed that the “singularity”- the point where technology surpasses human intelligence and control- has already arrived. “We are now, like, in the singularity. This is the moment,” he said.
The call for restraint is echoed by Anthropic, which has previously supported an intentional slowdown. Earlier this year, the US government invoked national security concerns to block the release of powerful models from both Anthropic and OpenAI; both companies complied until the orders were eventually reversed.
