Ethics concerns mount over Grok AI’s role in U.S. government

Elon Musk’s Department of Government Efficiency (DOGE) has quietly rolled out an adapted version of his Grok AI chatbot across federal agencies, raising alarms about potential privacy breaches and conflicts of interest. Reuters cited three sources inside DOGE saying the team has been using Grok to sift through and analyze sensitive government data, generating reports and insights at speeds beyond traditional methods. DOGE breaches ethics with the Grok AI move According to the three insiders, DOGE engineers installed custom parameters atop Grok, a chatbot that Musk’s xAI launched in late 2023, to accelerate data review and automate report writing. “They feed it government datasets, ask complex questions, and get instant summaries.” An Insider. Another insider added that DOGE staff had encouraged Department of Homeland Security employees to use Grok for internal analyses despite the tool lacking formal agency approval. What’s unclear is exactly what classified or personally identifiable information has been uploaded into Grok or how heavily it has been trained on federal records. If sensitive material were included, the practice could run afoul of federal privacy statutes and conflict-of-interest rules. Five ethics and technology experts warn that such access might give Musk’s companies disproportionate insights into non-public contracting data and even help refine Grok itself for private gain. In theory, any AI model trained on confidential government datasets must navigate strict legal safeguards. Data-sharing protocols typically involve multiple sign-offs and oversight to prevent unauthorized disclosure. Through sidestepping those checks, DOGE risks exposing millions of Americans’ personal details, and handing xAI a trove of real-world information unavailable to competitors. DOGE insists its mission is to root out waste, fraud, and abuse. A DHS spokesperson told Reuters that DOGE never pressured staff to adopt any specific tool. “We are focused on efficiency,” the spokesperson said. But two sources counter that, over recent weeks, DOGE representatives have pushed DHS divisions to pilot Grok for tasks ranging from immigration caseload analysis to budget forecasting, even after DHS abruptly blocked all commercial AI platforms over data-leak fears. Under current DHS policy, employees may use commercial chatbots only for unclassified, non-confidential work, while a bespoke DHS AI handles sensitive records. But when ChatGPT and others were disabled in May, DOGE’s advances occurred in a legal gray zone: the internal DHS bot remained live, but Grok was never formally on-boarded. Is Musk using DOGE to centralize control? Beyond DHS, DOGE’s reach extends into Department of Defence networks, where about a dozen analysts were reportedly informed that a third-party AI tool was monitoring their activity. Although DoD spokespeople have denied DOGE guided any AI deployments, departmental emails and text-message exchanges obtained by Reuters suggest otherwise. Critics see these moves as illustrative of Musk’s broader strategy to leverage AI to centralize control over bureaucracy and then monetize the resulting data flow. “There’s a clear appearance of self-dealing.” Richard Painter, a government ethics professor. If Musk directly ordered Grok’s deployment, he could be violating criminal statutes that bar officials from influencing decisions that benefit their private interests. At the heart of the debate is Grok’s dual role as a public-facing chatbot on X and an experimental analytics engine inside government firewalls. xAI’s website even hints that user interactions may be monitored “for specific business purposes,” suggesting that every federal query could feed back into Grok’s learning loop. Two DOGE staffers , Kyle Schutt and Edward Coristine, the latter known online as “Big Balls,” have spearheaded much of the AI initiative. While they declined to comment, their efforts fit a pattern; over the past year, DOGE has dismissed thousands of career officials, seized control of secure databases, and championed AI as the ultimate tool for bureaucratic overhaul. Privacy advocates warn that integrating unvetted AI into high-stakes national-security environments is a recipe for data leaks, identity theft, and foreign adversary exploitation. “This is about as serious a privacy threat as you get,” says Albert Fox Cahn of the Surveillance Technology Oversight Project. There are also fears that with little transparency and few guardrails, DOGE’s AI experiment could reshape federal data governance, whether the public realizes it or not. KEY Difference Wire : the secret tool crypto projects use to get guaranteed media coverage

Source: Cryptopolitan