# EFF Urges Strong AI Safeguards to Protect Constitutional Rights **Washington D.C.** – Governments must not adopt emerging and powerful artificial intelligence (AI) technologies without simultaneously implementing strong and clear safeguards to protect constitutional rights. This was the core message delivered by **Electronic Frontier Foundation (EFF)** Senior Policy Analyst **Dr. Matthew Guariglia** during his testimony today before the **House Homeland Security Subcommittee on Cybersecurity and Infrastructure Protection**. ## The Surveillance Threat of Generative AI During the hearing, titled "[The AI Security Landscape: How Frontier Models, Agentic AI, and AI Coding Tools Are Reshaping Cybersecurity and Critical Infrastructure Resilience](https://homeland.house.gov/hearing/the-ai-security-landscape-how-frontier-models-agentic-ai-and-ai-coding-tools-are-reshaping-cybersecurity-and-critical-infrastructure-resilience/)", Dr. Guariglia underscored the profound risks associated with governmental use of generative AI. He warned that deploying such technology for mass surveillance could "supercharge unconstitutional violations of civil liberties." ## The "Black Box" Problem and Critical Infrastructure Dr. Guariglia also highlighted the dual challenge of government secrecy and the proprietary "black box" nature of for-profit AI technologies. This combination, he explained, prevents both the public and lawmakers from understanding when AI models make critical errors. Such mistakes, he warned, could have serious implications for the cybersecurity of vital critical infrastructure and the lives of individuals. ## A Track Record of Costly Mistakes Illustrating his concerns, Dr. Guariglia pointed to concrete examples of AI failures. "AI also has a track record of getting things wrong—from false citations on legal briefs to a [major AI mistake](https://www.nbcnews.com/politics/immigration/ice-error-meant-recruits-sent-field-offices-proper-training-sources-sa-rcna254054) that sent **DHS** recruits to the field without proper training," he stated in his opening remarks. He suggested that many more consequential errors likely remain unknown due to classification, hindering a thorough accounting of AI's shortcomings. ## Reining in Agencies, Not Just AI Responding to a question from Subcommittee Ranking Member **Delia Ramirez**, D-Ill., Dr. Guariglia articulated the **EFF**'s broader perspective: "At this level the question is not how do we rein in AI, it’s how do we rein in the agencies that would unleash AI on the American public." This statement encapsulates the **EFF**'s focus on governmental accountability and the need for agencies to exercise restraint and implement robust oversight when considering AI deployment. Readers can access Dr. Guariglia's full prepared testimony [here](https://www.eff.org/document/06-04-2024-matthew-guariglias-prepared-testimony-house-subcommittee).