Project Description

The proliferation of publicly accessible generative AI (GenAI) tools has enabled the unprecedented spread of false information in textual, audio, and visual forms. As AI-generated media becomes increasingly difficult to discern from legitimate communications, governments are concerned that GenAI tools can be used to manipulate public sentiment. Already seen in propaganda movements amidst elections and the spread of fake news, the World Economic Forum has designated AI-powered misinformation as 2024's top immediate risk in undermining democratic processes.

The City of San Jose's interest in responding to the potential for GenAI to manipulate government communications led the San Jose Government team to produce an AI Misinformation Guidelines Manual and accompanying educational presentation for internal social media staff. This project recommends protocols that do not require extensive technical background to implement, addressing resource limitations within local governments that typically hinder timely responses toward emerging threats. Furthermore, San Jose is known for founding the GovAI Coalition, with over 300 agencies as members to promote standard setting for responsible AI use in government. The City of San Jose's leadership in this issue area can foster local-level collaboration to safeguard public communications and become a model for other local governments to look toward

Problems Tackled

Government regulations have not kept up with the advancement of free-to-use Generative AI tools within the last two years. With comprehensive federal laws still in the works, the responsibility of governance shifts to state and local governments. Though comparatively under-resourced, these lower levels of government can make decisions and respond in a more timely manner while also addressing specific concerns within the local community. The San Jose team acknowledges these different strengths by recommending both proactive and reactive strategies for safeguarding against Generative AI manipulations of official city communications. The AI Policy Manual presents strategies that can be easily integrated into existing digital communications and maintain applicability amidst the rapid development of GenAI tools.

Methodology

The team first conducted a literature review of academic findings, government responses, and local case studies to develop a preliminary understanding of prevailing methods in existing GenAI and digital misinformation responses. Within the literature review, the team concluded three key response categories: policy approaches addresssing mis/disinformation, technical tools for AI detection and response, and public education and media literacy initiatives to encourage community involvement. These themes were presented to Privacy and AI Analysts within the San José government and developed via biweekly meetings to convert findings into an actionable workflow including comprehensive incident documentation, authenticity verification of official government communications, and public awareness via a social media campaign.

Project Impact and Future Work

This project represents a pivotal step in safeguarding the integrity of government communications against the growing threats of AI and deepfake technologies. By developing comprehensive guidelines, policies, and educational materials, the Paragon San José Government team hopes the City of San Jose will establish robust defenses against misinformation and ensure that residents can trust official communications. The project's deliverables, including a literature review, policy manual, and educational presentations, will become essential tools for training public information officers and educating the public. The deliverables not only provide communications recommendations resilient to continuous AI developments, they also set precedence for other municipalities by emphasizing the necessity and tangible strategies for local-level response.

Contributors