Executives at the artificial intelligence company Anthropic are scheduled to meet with senior White House officials amid renewed national security worries regarding the firm's most recent product launch.
The meeting is planned for Monday in Washington, D.C., between Anthropic representatives and the U.S. Department of Commerce, an agency headed by Secretary Howard Lutnick, according to two individuals familiar with the situation.
This follows Anthropic's decision on Friday to block all public access to the recent release of its newest AI tool, which the company has previously described as "too powerful."
The firm took this step after the U.S. government prohibited Anthropic from allowing any foreign nationals to access the technology.
The AI tool in question is called Fable 5 or Mythos 5. Fable 5 is a version of the tool with additional safeguards that was made available to the public, whereas Mythos 5 features different controls and is accessible only to a specific set of organizations.
Both represent a new iteration of Claude Mythos, an Anthropic AI model that generated significant attention when the company initially provided preview and testing access to a relatively limited number of organizations in April, including departments within the U.S. government.
Anthropic CEO Dario Amodei is expected to attend the meeting with Secretary Lutnick, according to a source.
A White House spokesperson declined to comment. Representatives from both the Commerce Department and Anthropic did not respond to requests for comment.
By releasing a version of Mythos to the public, Anthropic stated last week that doing so "comes with risks."
"Fable's capabilities exceed those of any model we've ever made generally available," the company added.
Within days of the release, the U.S. government said it had "become aware" of a potential "jailbreak," meaning an opportunity for someone to make an AI tool perform actions it was not intended or designed to do.
Anthropic said on Friday that it had only received "verbal evidence" of the alleged jailbreak vulnerability.
The rift between Anthropic and the government is the latest this year, following Anthropic's lawsuit against the U.S. Department of Defense over a dispute about how its models can be utilized.
However, tensions appeared to be easing a few weeks ago when the company met with senior White House officials in what was described as a "productive" session.
The Monday meeting with the Department of Commerce is expected to involve more documentation of the alleged issue, according to one of the sources.
It remains unclear, however, whether Anthropic will be able to restore access to Fable 5 and Mythos 5 after the meeting.
Dozens of technology leaders and executives in the cybersecurity field have urged the U.S. government to allow Anthropic to make the models available to the public again.
In an open letter, security personnel from Nvidia, Zoom, and Mercedes-Benz, along with former security staff from the U.S. government and Google, called on Lutnick to lift the restrictions placed on Fable 5 and Mythos 5.
They also requested that the government "commit to an open, scientific and transparent process of handling AI risk assessments in the future."
The White House has signaled a relatively hands-off approach to regulating AI and has even expressed interest in benefiting financially from it.
Yet the recent action against Anthropic has raised concerns among AI developers and security experts.
"To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous," the signatories added.