Has "Don't Be Evil" Become a Thing of the Past? Alphabet Inc. Class C (GOOGL.US) Refuses to Follow in Anthropic's Footsteps, Signs Confidential AI Collaboration Agreement with the US Department of Defense.

According to media reports citing informed sources, Alphabet Inc. Class C (GOOGL.US) has joined the ranks of more and more technology companies by signing an agreement with the US Department of Defense, allowing its artificial intelligence (AI) models to be used for classified work. The report states that the agreement allows the US Department of Defense to use Alphabet Inc. Class C's AI models for "any lawful government purpose." This puts Alphabet Inc. Class C in the same category as OpenAI and Musk's xAI, both of which have also signed agreements to provide AI models for classified purposes. Classified networks are used for handling a wide range of highly sensitive work, including mission planning and weapon targeting. According to reports, the agreement between Alphabet Inc. Class C and the US Department of Defense requires the company to assist in adjusting the security settings and filters of its AI model as requested by the government. The contract includes a provision that states: "Both parties agree that AI systems may not be used for domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight and control." However, the report also notes that the agreement stipulates that Alphabet Inc. Class C does not have the authority to control or veto lawful government action decisions. Alphabet Inc. Class C stated that it supports government agencies in carrying out classified and non-classified projects. A spokesperson stated that Alphabet Inc. Class C still adheres to industry consensus that AI should not be used for domestic mass surveillance or deployed in autonomous weapons without proper human supervision. The spokesperson for Alphabet Inc. Class C stated, "We believe that offering access to commercial models via API with industry standard practices and terms (including running on Alphabet Inc. Class C infrastructure) is a responsible way to support national security." The US Department of Defense stated that it has no intention of using AI for mass surveillance of the American people or developing weapons that can operate without human involvement, but hopes to allow AI for "any lawful purpose." US Department of Defense advocates for "All Lawful Use" standard as Anthropic refuses to compromise and is blacklisted In 2025, the US Department of Defense signed agreements worth up to $200 million each with several major AI labs, including Anthropic, OpenAI, and Alphabet Inc. Class C. Prior reports suggested that the US Department of Defense has been pushing top AI companies like OpenAI and Anthropic to enable their tools to operate on classified networks and to lift standard restrictions usually imposed by these companies on users. The US Department of Defense's stance on its procurement standard of "All Lawful Use" is becoming increasingly firm. This standard grants military agencies the right to use suppliers' AI capabilities within the limits allowed by existing US laws and military policies, without being subject to additional ethical restrictions imposed by tech companies. In the implementation of this standard, Anthropic, a major competitor of Alphabet Inc. Class C, has faced severe suppression by the US government. Anthropic has refused to compromise and insisted on maintaining restrictions in its contracts prohibiting the use of the Claude model for domestic mass surveillance and completely autonomous weapons without human supervision. As a result, Anthropic has been labeled a "supply chain risk" by the US government and given a six-month transition period to fully withdraw from US government networks. US President Trump even ordered all federal agencies to immediately cease using Claude, accusing the company of attempting to force the military to adopt its values as a "radical leftist" company. On March 9, Anthropic filed a lawsuit against the US Department of Defense and other federal agencies for listing it as a "supply chain risk." On March 26, Judge Rita F. Lin of the Northern District of California Federal District Court approved Anthropic's request for a temporary injunction, preventing the US federal government's order to ban the use of Anthropic's AI technology from taking effect. It is worth noting that tensions between Anthropic and the US government appear to be easing. Trump stated in an interview last Tuesday that he hopes to reach an agreement to allow Anthropic's AI models to be used within the US Department of Defense. Earlier this month, Anthropic released the Mythos model. Due to its advanced cybersecurity capabilities, the model is only being offered to selected companies. Anthropic stated that it is in "ongoing discussions" with US government officials about the model. Despite the fact that the model was launched at the height of the dispute between Anthropic and the Department of Defense, its release has opened the door for the company to regain favor with the government. "Don't be evil" becomes a thing of the past? If the reported agreement in the above article is true, it will mark a fundamental shift in Alphabet Inc. Class C's corporate strategy after a long-term struggle between national security and ethical guidelines. Faced with Microsoft Corporation (MSFT.US)Azure and Amazon.com, Inc. (AMZN.US)AWS's long-standing dominance in the government classified cloud market, Alphabet Inc. Class C urgently needs to make strategic breakthroughs in its business through such defense-grade AI integration projects. The confidential AI agreement reached with the US Department of Defense by Alphabet Inc. Class C is in fact the inevitable result of profound revisions made to its internal AI ethics guidelines over the years. In 2018, under strong employee protests, Alphabet Inc. Class C withdrew from the Maven project, which involved using AI to analyze drone footage, and subsequently established clear AI principles, promising not to develop monitoring technology for weapons or technologies likely to cause harm. However, with the intensification of the global AI arms race, Alphabet Inc. Class C's stance has fundamentally changed. In February 2025, Alphabet Inc. Class C quietly revised its internal AI principles, directly removing the section titled "Applications we will not pursue," thus eliminating the commitment not to use AI for weapons and surveillance. At an internal all-hands meeting in March 2026, Alphabet Inc. Class C's Vice President of Global Affairs, Tom Lue, sent a clear signal to employees concerned about defense contracts, stating that the company is "more leaning towards" national security work and believes these contracts are in line with the updated AI guidelines. However, internal dissent has not been quelled. Over 100 Alphabet Inc. Class C AI employees recently sent a joint letter to Chief Scientist Jeff Dean demanding that the company prohibit the military from using Gemini for domestic surveillance or autonomous weapons, highlighting the significant tension between corporate business goals and employee ethical principles. Anthropic's departure has created a strategic vacuum for Alphabet Inc. Class C and OpenAI. Faced with the US military's strong stance, Alphabet Inc. Class C has chosen to accept the overall framework of "All Lawful Use" and is attempting to mediate by "proposing" clauses in specific contract texts to prevent the misuse of mass surveillance and uncontrolled autonomous weapons. Analysts point out that this strategy satisfies the Pentagon's bottom line demand for technological autonomy while attempting to maintain a corporate ethical image on the surface, reflecting concessions in the boundaries of enterprise in the face of political competition pressures. The agreement reached between Alphabet Inc. Class C and the US Department of Defense indicates that this tech giant is seeking to solidify its position as a core supplier in the next generation "defense-AI-industrial complex." In this process, traditional corporate ethical boundaries inevitably compromise to meet the absolute need for national security strategizing. (Note: Some parts may have been omitted for brevity)