Anthropic has appointed national security veteran Richard Fontaine to its Long-Term Benefit Trust, signaling a deeper focus on AI's security implications as it also rolls out new AI models designed for government defense work. The twin moves underscore the AI firm's strategy to navigate the complex intersection of advanced AI development and national security demands.

• Guarding the guardians: Anthropic's Long-Term Benefit Trust, a distinct governance body, aims to steer the company's work toward safety over pure profit, and it has the power to appoint some board members. Fontaine, CEO of the Center for a New American Security (CNAS), joins other trustees like Zachary Robinson of the Centre for Effective Altruism and Neil Buddy Shah from the Clinton Health Access Initiative, according to TechCrunch. Anthropic CEO Dario Amodei stated Fontaine's "expertise comes at a critical time as advanced AI capabilities increasingly intersect with national security considerations."

• Claude goes classified: Fontaine’s appointment coincided closely with Anthropic’s introduction of "Claude Gov," a suite of AI models built for U.S. defense and intelligence agencies operating in classified settings. These systems are designed for tasks such as strategic planning and improved handling of sensitive data, with Anthropic noting the models "refuse less" with such information. This isn't Anthropic's first venture into defense; last November, it partnered with Palantir and AWS to offer its AI for defense applications.

Anthropic’s moves mirror a broader industry pattern, with other major AI developers also actively engaging the defense and intelligence sectors. OpenAI is working to build stronger connections with the U.S. Defense Department, Meta is making its Llama models available for similar government uses, and Google is developing a version of its Gemini AI for classified environments, as reported by Ars Technica.

• More from the Anthropic files: Beyond board appointments, CEO Dario Amodei has been vocal, warning that AI could displace up to half of entry-level white-collar jobs in five years and potentially push unemployment to 20%. He also claimed Anthropic's AI hallucinates less than humans, though some believe he might be understating AI's disruptive job impact.