Written by: Techub News

Recently, the tense relationship between AI star company Anthropic and the U.S. Department of Defense has become the focal point of the intersection of technology and politics. The Defense Department has threatened Anthropic under the banner of "supply chain risk," while the company has staunchly refused to make concessions on two key applications. In this much-watched controversy, Anthropic co-founder and CEO Dario Amadei gave an exclusive interview to "Face the Nation," fully articulating the company's stance, negotiation details, and profound reflections on the balance between AI ethics and national security for the first time.

“Red Lines”: Mass Surveillance and Fully Autonomous Weapons

Dario Amadei directly pointed out that Anthropic is not unwilling to cooperate with the government; on the contrary, the company has been among the "most proactive" AI enterprises. They were the first to deploy models on classified clouds and the first to customize models for national security purposes, with their technology broadly applied in the intelligence community and military to support cybersecurity, operational support, and various tasks.

However, two clear "red lines" exist for cooperation. The first is the prohibition of using AI for "domestic mass surveillance." Amadei explained that AI technology is developing at an unprecedented speed, making some surveillance methods that were previously permitted legally but technically unfeasible now possible. For example, the government could purchase vast amounts of data collected by private companies (including location, personal information, political preferences, etc.) and conduct massive analyses using AI. This behavior might currently not be illegal but has only become "useful" due to the advent of AI. Amadei believes that the pace of technological advancement has outstripped the pace of legal updates, presenting a risk of surveillance methods "staying ahead of the law," which touches upon the fundamental right of citizens not to be monitored by the government.

The second red line is the prohibition of developing "fully autonomous weapon systems." Amadei clarified that this does not refer to the "semi-autonomous weapons" currently used in Ukraine or potentially in Taiwan, but to "weapons that fire without any human involvement." He provided two core reasons for opposition:

• Insufficient technical reliability: Today's AI systems are nowhere near reliable enough to support fully autonomous weapons. Anyone who has worked with AI models understands that they have a "fundamental unpredictability," which has not yet been resolved on a purely technical level.
• Lack of oversight and accountability: If an army composed of drones or robots operates without human supervision, who decides the attack targets? Who bears the responsibility for decision-making? There has been no substantial discussion in society about the oversight framework for such weapons.

Amadei emphasized that Anthropic is not absolutely against fully autonomous weapons. He acknowledged that democratic countries might need them for defense due to adversaries potentially possessing such weapons in the future. But the premise is that the technology must be reliable enough, and society must establish corresponding oversight mechanisms. Anthropic has even offered to collaborate with the Defense Department to co-develop and test relevant technological prototypes in a "sandbox" environment, but the other party demanded to be able to "do as they please" from the start, leading to the inability to advance cooperation.

Negotiation Breakdown: Three-Day Ultimatum and “Punitive” Measures

Amadei revealed that the timeline of the entire incident was very urgent and completely controlled by the Defense Department. The other side issued a three-day ultimatum, demanding that Anthropic either agree to its terms or risk being classified as "a supply chain risk" or face measures related to the Defense Production Act.

During this time, there were several exchanges of communication. Amadei pointed out that the Defense Department had sent some texts that seemed to meet Anthropic’s requirements, but they included vague wording such as "if the Pentagon thinks it is appropriate" or "acting within the law," which did not make any substantive concessions. A Defense Department spokesperson also reiterated their position on Twitter: "Only all lawful uses are permitted," consistent with the terms given to Anthropic. Amadei believed that the other side had never genuinely accepted the company's exception requests.

Subsequently, the Defense Secretary publicly labeled Anthropic's actions as "selfish" and accused it of "putting American lives at risk, endangering our troops, and jeopardizing our national security." In response, Amadei stated that even when facing unprecedented extreme measures from the government (classifying an American company as a supply chain risk, which is usually only used against foreign adversaries), Anthropic remained committed to supporting the Defense Department in ensuring the continuity of technical support until transitioning to other willing suppliers.

Amadei expressed deep concerns about service disruption. Frontline officers he contacted indicated that losing Anthropic's technical support would set related work back "six months, twelve months, or even longer." He emphasized that the company had been striving to reach an agreement, but the entire process's urgency and confrontational nature were driven by the Defense Department.

Principle Dispute: Private Company vs. Government Authority

The interviewer asked sharply: why should a private company (Anthropic) have more authority than the Pentagon itself to decide how AI is used in the military?

Amadei provided a multi-layered response:

First, actual impact is minimal. He claimed that to their knowledge, frontline personnel had never encountered the applications restricted by these two "red lines" in actual work. These prohibited use cases account for only 1% of the Defense Department's demands, while the remaining 99% of applications (such as intelligence analysis, logistics optimization, cybersecurity) are progressing smoothly. The company has engaged in extensive collaboration with the Defense Department and other government departments without encountering any issues.

Second, technological novelty and legal lag. Amadei acknowledged that, in the long run, this should not be a dispute between private companies and the Pentagon, but rather the responsibility of Congress. When the pace of technology (such as AI analyzing vast personal data) far outstrips the updates of the law (such as judicial interpretations of the Fourth Amendment or congressional legislation), legislative intervention is needed. However, "Congress is not the fastest-moving institution in the world." Before Congress catches up, as developers at the forefront of technology, Anthropic has the responsibility to recognize the capabilities and risks of technology and draw boundaries when legal intentions may be circumvented by technology.

Third, free market and choice. Amadei pointed out that in a free market, different companies can offer different products based on different principles. If the Defense Department dislikes Anthropic's service terms or principles, they are free to choose other suppliers. This should be the normal way to address such disputes. However, the government’s actions have gone beyond this scope: they have not only extended the controversy to other government departments beyond the Defense Department, attempting to punishingly revoke Anthropic's other contracts, but also invoked the "supply chain risk" classification. This classification means that any private company with a contract relationship with the military cannot use Anthropic's technology in its military contract-related business. Amadei believes that this clearly intervenes in private enterprise behavior, and its "punitive" intent is hard to explain otherwise. To his knowledge, the "supply chain risk" classification has never been applied to an American company and was previously only directed at foreign entities like the Russian Kaspersky Lab.

Values and Survival: Patriotism, Neutrality, and Business Resilience

When asked about former President Trump's labeling of Anthropic as a "left-wing woke company" and whether this decision was ideologically driven, Amadei firmly denied it. He stated that the company strives to remain neutral, only voicing opinions in the field of AI policy where it has expertise and not engaging in general political issues. He cited examples of attending events with the president and senators discussing AI energy issues and supporting most of the government's AI action plan to demonstrate the company's "prudent fairness."

Regarding the company's survival ability, Amadei showed strong confidence. He clarified that the Defense Secretary's tweet contained inaccuracies, exaggerating the impact range of the "supply chain risk" classification. The actual legal stipulations are much more limited: they only prohibit other companies from using Anthropic's technology in the course of fulfilling their military contracts. Therefore, he believes the actual business impact of this classification is "quite small." He accused the relevant tweet of attempting to create "fear, uncertainty, and doubt," but the company would not let that succeed, asserting that "we will not only survive but will thrive."

At the end of the interview, Amadei was asked whether he thought the government's actions were "an abuse of power." He did not affirm directly but repeatedly emphasized that this action was "unprecedented," and pointed out that it could be clearly seen from some government statements and wording that it was "retaliatory and punitive." As for legal action, he stated that they had only received tweets for now and had not yet received any formal official notification of action. Once received, the company would closely examine it and "challenge it in court."

Amadei concluded that everything Anthropic does is out of patriotism, aimed at supporting U.S. national security. They draw red lines because they believe crossing these lines would go against American values. When the government intervenes in private economy in unprecedented ways, they exercise the right granted by the First Amendment to express dissent. “Dissenting from the government is the most American thing in the world,” he said, “In everything we do, we are patriots. We defend the values of this country.”

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。