The White House plans to grant federal agencies access to Anthropic's powerful AI model, Mythos.
```
The US government is advancing plans to bring a powerful new model from Anthropic, Mythos, into major federal agencies in a specific version, but only after cybersecurity safeguards are put in place to reduce concerns about the tool potentially significantly amplifying cyberattack risks.
According to Bloomberg, White House Office of Management and Budget (OMB) federal Chief Information Officer Gregory Barbaccia said in an email to cabinet department officials on Tuesday that OMB is setting up protective measures allowing agencies to begin using this “strictly controlled” AI tool, and noted more information will be provided “in the coming weeks.”
So far, Mythos has only been open to a limited group of tech companies, financial institutions, etc. Anthropic previously stated that this move is to prevent hackers from weaponizing the model’s capabilities. Against this backdrop of strict controls, OMB’s push for government agencies to access Mythos has attracted great attention to the potential risks.
For the market, this move highlights the government’s tug-of-war between accelerating adoption of advanced AI and controlling the spillover risk of dual-use offensive and defensive capabilities. Investors will focus on the potential window for federal agencies’ demand for high-capability models, and on the repricing of cybersecurity risk assessment and protection investments resulting from this.
OMB deploys protection mechanisms, timeline remains unclear
In an email titled “Mythos Model Access,” Barbaccia wrote:
“We are closely collaborating with model providers, other industry partners, and the intelligence community to ensure that before releasing a modified version of the model to agencies, appropriate guardrails and security measures have been put in place.”
The email does not specify whether agencies will ultimately be granted access to Mythos, nor does it provide specific timelines or usage guidelines. The email only informs technical and cybersecurity leaders at the agencies that relevant details will be disclosed in the “coming weeks.”
A White House official, responding to Bloomberg, said the government will continue to work with AI companies to ensure their models help fix critical software vulnerabilities, but did not respond to specific issues.
Cybersecurity risks drive Anthropic’s strict control of release
The core reason Mythos’ release has been strictly limited is because it possesses dual-use capabilities for both cyber offense and defense, which has caused high alert across the industry and government.
According to an Anthropic official speaking to Bloomberg, before Mythos was officially released to a limited group, Anthropic had already provided comprehensive briefings on its capabilities to senior officials in multiple US government departments, covering application scenarios in both cyber offense and defense.
Agencies involved in the briefings include the Cybersecurity and Infrastructure Security Agency (CISA) and the Center for AI Standards and Innovation (CAISIT), among others. The source said that Anthropic has continued to cooperate with the government on security issues raised by the model.
Institutions that have gained access to Mythos so far are required to use it solely to assess their own cybersecurity risks. This targeted access strategy reflects Anthropic’s cautious balance between commercial promotion and risk management.
Risk DisclaimerThe market has risks, and investment needs caution. This article does not constitute personal investment advice, nor does it take into account any individual user’s special investment objectives, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their specific situation. Investment based on this information is at your own risk. ```