Axios, which was first to report the news, said Friday that the meeting is meant to resolve the artificial intelligence (AI) company’s dispute with the government at a time when the company’s Mythos model is causing widespread concerns about cybersecurity.

The Pentagon banned the use of Anthropic’s products after a dispute over how those products can be used, and Anthropic is currently suing the Pentagon over that ban.

However, Axios quoted a source close to the negotiations as saying that the government would be “grossly irresponsible” if it didn’t access the capabilities of Anthropic’s new model.

The Wall Street Journal reported on the scheduled meeting Friday, saying that Amodei wants to ease tensions with the Trump administration and that Anthropic has been briefing government officials about Mythos and is in talks to give agencies early access to the model.

The potential risks around Mythos led both sides to seek to make peace, according to the WSJ report. Amodei has already participated in meetings in which tech and financial executives discussed preparation for Mythos with Vice President JD Vance and Treasury Secretary Scott Bessent, per the report.

CNBC also reported Friday that it confirmed the planned meeting between Amodei and Wiles, saying that the meeting will focus on Mythos and represents a thaw in the nearly two-month standoff between the government and Anthropic.

Reuters reported Friday that a White House official said the administration works with government and industry, including AI companies, to ensure the security of software.

It was reported Thursday (April 16) that Anthropic has also met with European Union officials to discuss concerns about Mythos and that more such meetings are planned.

It was also reported Thursday that Anthropic is ready to begin offering Mythos to British banks. This move marks an expansion of the AI company’s Project Glasswing, which offers select organizations early access to the AI model.

Anthropic launched the program after learning of Mythos’ ability to spot and possibly exploit weaknesses in cyber defenses.