TLDR: Delaware’s Department of Technology and Information has implemented a new policy for state employees regarding the use of generative AI. The policy approves the use of tools like ChatGPT and Microsoft Copilot but bans Chinese-developed AI models such as DeepSeek, labeling them as ‘malicious’ due to national security risks. This decision reflects a larger trend of state-level actions influenced by geopolitical competition with China, turning technology procurement into a form of foreign policy.
Delaware’s Department of Technology and Information recently enacted a new policy governing the use of generative AI by state employees, a move that, on the surface, appears to be a routine administrative update. The policy greenlights popular tools like ChatGPT and Microsoft Copilot but takes the significant step of explicitly banning specific Chinese-developed AI models, labeling them as ‘malicious.’ While it may seem like a localized decision, this policy is one of the clearest signals to date that geopolitical risk has evolved from a theoretical concern into a primary driver of day-to-day government technology regulation. For policymakers, AI ethicists, and public affairs specialists, this development is a critical inflection point, demanding a re-evaluation of long-term strategies for creating balanced and globally interoperable AI governance frameworks.
From Statehouse to World Stage: When Procurement Becomes Foreign Policy
The decision by Delaware is not happening in a vacuum. It mirrors actions taken by other states, including Texas, New York, and Virginia, which have also moved to prohibit the use of AI applications like DeepSeek on government devices, citing national security concerns. This pattern suggests the emergence of a bottom-up, state-level implementation of a broader, national-level anxiety regarding technological competition with China. What was once the domain of federal export controls and intelligence agency warnings is now manifesting in state IT procurement lists. This shift transforms routine administrative choices into acts of foreign policy, establishing a de facto list of approved and unapproved nations for AI development. For government advisors and regulators, this means that technology stacks are no longer neutral; they are statements of geopolitical alignment. The key challenge is now to manage this reality without stifling innovation or creating an unwieldy patchwork of conflicting state and federal rules.
The ‘Malicious Model’ Distinction: A New Front in the Tech Cold War
Delaware’s policy introduces the potent designation of a ‘malicious AI model,’ a term that appears to be defined not by technical fallibility but by national origin and the potential for data exfiltration to a foreign adversary. By explicitly permitting Western-developed enterprise tools like Microsoft Copilot and Amazon Bedrock—which offer contractual data protections—while banning Chinese counterparts like DeepSeek and RedNote, the policy draws a clear line in the sand. This creates a bifurcated AI ecosystem: a ‘trusted’ ecosystem of tools from the U.S. and its allies, and an ‘adversarial’ one. This dynamic is a core tenet of the escalating U.S.-China tech rivalry, where leadership is not just about having the best technology, but about controlling the underlying infrastructure and data flows. As AI becomes more deeply embedded in government operations, from analyzing sensitive constituent data to managing critical infrastructure, the provenance of these tools becomes a paramount security concern. This forces ethics professionals and policymakers to grapple with a difficult question: how do you verify and trust an AI model when its training data and inner workings are intrinsically linked to a nation-state with competing geopolitical interests?
The Interoperability Dilemma: Navigating a Fractured Global AI Landscape
Perhaps the most significant long-term implication of policies like Delaware’s is the threat it poses to the dream of a globally interoperable AI framework. For years, ethicists and international bodies have worked towards creating universal standards for AI safety, fairness, and transparency. However, the rise of politically motivated AI bans accelerates the formation of a technological ‘splinternet,’ where different AI ecosystems operate under mutually exclusive rules and values. This complicates international research collaboration, undermines efforts to create global safety benchmarks, and creates immense challenges for multinational organizations that must navigate these diverging compliance regimes. A fragmented landscape risks creating isolated pockets of AI development, potentially slowing down progress and making it harder to address global challenges like climate change or pandemic response, which benefit from open collaboration. The critical task for policy leaders is to find a way to mitigate legitimate security risks without completely sacrificing the benefits of an open and interconnected global AI community. This requires a nuanced strategy that goes beyond simple bans and fosters deeper diplomatic engagement on technology standards.
A New Mandate for Geopolitical Literacy in Tech Policy
Delaware’s AI policy is a stark reminder that technology and geopolitics are now inextricably linked. What may have been dismissed as a niche concern is now a practical reality influencing decisions in state capitals across the country. For the government, policy, and ethics professionals shaping our technological future, this marks a point of no return. The core takeaway is that geopolitical risk assessment can no longer be an afterthought in technology governance; it must be a central pillar. Looking ahead, the challenge will be to craft policies that are both secure and strategically sound, ensuring the United States can protect its interests while continuing to lead the global conversation on responsible and ethical AI innovation. The next critical step is to watch how federal agencies harmonize these state-level actions into a coherent national strategy that balances defensiveness with a proactive vision for AI leadership on the world stage.
Also Read:
