Project Description

The proliferation of AI-generated deepfakes and misinformation through high traffic media channels and phishing schemes are a significant threat to civic life across America. Particularly, deceptive content can erode resident-government relations, trust in democratic processes, and the safety of public and private settings across the United States.

For Fort Collins specifically, a misleading video clip that appeared to show a police officer using extreme force to subdue a community member garnered viral outrage online and even lawsuits – raising policymakers’ concerns about the increased threats AI-generated content could pose to individuals’ reputations. Government leaders in Fort Collins would not only like to protect public officials from such harms, but also non-government affiliated community members from reputational damage, while ensuring citizens do not believe falsified information that could disseminate in the new era of AI-generated content. Among these are small, local businesses who lack the technical infrastructure to handle attacks and vulnerable age groups who may be too young or too old to have come across exposure to deepfake-related risks. The city of Fort Collins is aiming for a practical, feasible, and city-ready strategy to (1) better inform citizens about the dangers of AI deepfakes, (2) integrate effective mitigation techniques in the most vulnerable channels, and (3) and create effective policy that aligns with city goals and will remain sustainable as generative AI evolves.

Methodology

The Fort Collins team began this initiative by conducting a holistic research program on the current state of AI Deepfakes, particularly focusing on four main fronts: (1) current regulations and policies surrounding AI technology in public input, (2) AI-driven risks in public input, (3) current technological solutions and detection tools for AI-generated content, and (4) AI literacy and training. After garnering a better understanding of the state of deepfakes across the country, we then narrowed down the specific concerns and threats posed to the City of Fort Collins through highly impactful conversations with the city’s Chief Information Officer, Kevin Wilkins, and the city’s Director of Technology, Jeff Willard. Through their Fort Collins specific insights, the team began developing a robust, 3-pillar strategic advisory covering (1) a public awareness campaign tailored to vulnerable Fort Collins channels, (2) technology solutions to be integrated into existing public and private sector organizations, and (3) policy advice for the legislation surrounding deepfakes. We finally conducted a suite of stakeholder interviews with Fort Collins groups and external experts, including forensic analysts at the Fort Collins Police Department, cybersecurity fellows at Stanford University, and social media researchers at the University of Wisconsin to support our research and outline future concerns and areas of AI deepfakes.

Project Impact and Future Work

This three-pillar approach covers an exhaustive framework that offers the City of Fort Collins a practical roadmap for strengthening their resilience toward AI deepfake threats, ultimately protecting citizens, public officials, and organizations from misinformation and scams. And as deepfakes know no barriers, this approach can act as a national model, advancing the resilience against AI-driven misinformation across the country. Specifically, the multi-lateral and multi-layered framework, consisting of policy awareness campaigns, integrated deepfake technology solutions, and proactive policy measures, applies to any mid-sized city that is looking to combine education, technology, and governance to proactively address the dangers that deepfakes present. But more than that, this framework is one that is tailored to the most challenging attribute that deepfakes pose: adaptability. By emphasizing scalability, partnerships, and data-driven iteration on technical and non-technical fronts, this framework creates an infrastructure that can adapt to the evolving nature of deepfake technologies and stand the test of time.

As for next steps, the city of Fort Collins can begin to apply these frameworks for their immediate concerns. The team prioritizes launching public awareness campaigns to enhance AI and deepfake literacy among citizens of Fort Collins, serving as a first-step to prevent individuals from being misguided and enhancing the trust between citizens and public officials (which was of major concern to the city). We also advise enhancing citizen engagement through a community-driven detection approach: specifically, by developing a government-run portal, preferably attached to the city website or integrated within its social media channels, for reporting suspected deepfakes. Additionally, the city could shortlist high risk local businesses for possible pilots of technology integrations. This could include equipping digital channels such as organizational emails and messaging systems, social media sites, and even susceptible brick-and-mortar technologies like video surveillance with the deepfake mitigation tools such as public APIs, watermarking, and blockchain encryptions. More long term actions include the incorporation of standardized, city-wide technology solutions and forming a dedicated deepfake task force to review and adapt policy to meet the needs for deepfake strategy for both public and private sector groups.

Contributors