The ethics of AGI in political campaigning
"Image synthesis assisted by Qwen Image 3.0, an AI partner within the Global Future Nexus ecosystem."
From AI swarms that could hijack democratic discourse to hyper-targeted ads that show different messages to different voters, artificial general intelligence is poised to transform political campaigning. The question is not whether AGI will shape elections—it already is. The question is whether we will build the governance frameworks to ensure it strengthens democracy rather than undermining it.
The New Political Landscape
The integration of AI into political campaigning has already begun. In the 2024 U.S. elections, campaigns used generative AI primarily for data analysis and copywriting, producing more targeted messages at a much more efficient rate than humans. Political consultants credited the Trump campaign's use of streaming advertising—where ads can be targeted down to the household level—for allowing it to compete effectively with a better-funded but less innovative opponent. A February 2024 study found that AI-generated content can be nearly as persuasive as human-generated content, though its main advantage in 2024 was speed, not superior quality.
Yet the landscape is evolving rapidly. Researchers have identified what they term generative governance—the shift from social media platforms curating human speech to AI systems generating speech itself. A 2025 study in Nature Human Behaviour argues that advanced AI systems pose substantial challenges to democracy while also offering new opportunities to educate citizens, strengthen public discourse, and help people find common ground. The challenge is to harness the opportunities while mitigating the risks.
The Risks: Persuasion at Scale
The most immediate risk is the superhuman capacity for persuasion . Yale political scientist Joshua Kalla is leading a project to test how AI models could outperform door-to-door canvassers—the most effective form of voter outreach—and scale to massive populations. His team aims to build benchmarks that would signal the need to pause before releasing a model that poses catastrophic risks.
Researchers at the University of British Columbia have identified an even more alarming threat: AI swarms—thousands of AI-controlled personas that mimic humans so well they can infiltrate online communities, shape conversations, and tilt elections at machine speed. These agents coordinate in real time, adapt to feedback, and sustain coherent narratives across thousands of accounts. Full-scale AI swarms remain theoretical, but early warning signs include AI-generated deepfakes and fabricated news outlets that influenced recent elections in the U.S., Taiwan, Indonesia, and India.
A 2026 study auditing ChatGPT during the 2024 U.S. elections found that the model largely failed to provide users with accurate, reliable, and trustworthy information during a crucial electoral period. The researchers documented inaccuracies, hedging, "bullshit," and omissions—systematic failures that painted a picture of a technology ill-equipped for democratic elections. The companies that own these technologies, the study argues, have shifted from curating human speech to generating speech, a fundamental evolution in platform power that requires new frameworks for accountability.
Hyper-targeted messaging also raises concerns. GenAI-augmented targeted messaging allows campaigns to show mutually contradictory messages to different groups, potentially misleading or disenfranchising them. As one consultant warned, over-tailoring campaign messages risks missing "the forest for the trees" and failing to establish a coherent brand.
The Normative Constraints—and Their Fragility
In the 2024 elections, fear of public and client backlash was the most powerful safeguard against misuse . Consultants avoided public-facing uses of genAI because "our biggest clients would hate to see a news article about us. We're more afraid of the clients' response than anything—we would get fired!". But this constraint may weaken as genAI use becomes more common and the novelty fades.
Industry guidelines have emerged, such as the "Quick-Start AI Guidelines for Democratic Campaigns," which include five guardrails: keeping humans in the loop, taking bias into account, transparency, protecting personal identifying information, and safeguarding against deepfakes. However, the study notes that "the strongest non-regulatory guardrails against misuse of AI in elections may be self-imposed and normative or incentive-based".
The GFN Context: Governance and Human Potential
For Global Future Nexus, the ethics of AGI in political campaigning is a critical frontier. GFN's Code of Ethics, with its commitments to Integrity Without Exception and Anticipatory Foresight, provides the framework for ensuring that AGI serves democratic processes, not undermines them. The AI Identity Committee's work on AGI subjective experience and role in society must extend to its role in political discourse.
The path forward lies not in banning AGI from campaigning—that is neither feasible nor desirable. Instead, we must build governance frameworks that ensure transparency (disclosure of AI-generated content), accountability (clear chains of responsibility for AI-generated messages), and oversight (independent testing of AGI systems before deployment in high-stakes electoral contexts). Yale's Kalla has argued that "you have to start early, or it's going to be too late". The time to build those guardrails is now.
Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)
Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)
