Tests of six widely used AI models found safeguards against election misinformation were easy to bypass, producing realistic image prompts and strategic guidance. The authors outline practical steps for companies, lawmakers, and civil society to reduce harms before November.
Six popular AI systems (ChatGPT, Gemini, Grok, Meta AI, Runway, Flux.2) were tested and produced material that could enable large-scale election misinformation campaigns.
Companies, lawmakers, and civil society need immediate, concrete actions—improved model safeguards, coordinated legal and policy measures, and resourced public defenses—to blunt AI-driven disinformation before November.
Testing the Safeguards Before November By Lawrence Norden and David Evan Harris Published on September 18, 2026 The 2024 U.S. presidential election was supposed to be the "first AI election." Experts warned that generative AI could flood voters with deepfakes and fabricated evidence of fraud, U.S. intelligence officials declassified intel reports on foreign influence operations already experimenting with the technology, and leading AI companies pledged to combat deceptive election content.
Such campaigns often merely echo existing fractures and arguments appearing in the United States. President Trump and his allies are engaged in a concerted campaign to undermine U.S. While the evolution of AI comes with great risks in the wrong hands, the good news is that there are steps AI companies, policymakers, and civil society can take immediately (and in some cases are already taking).
Testing the Safeguards Before November Just Security. Two years later, the picture looks very different. AI tools have become far more sophisticated and widely available.
This essay was written with Nathan E. Sanders, and originally appeared in The Guardian . There are plenty of signs that AI will make all of our experiences of the US midterm elections worse. Voters have anxiety about AI’s impacts on the country. Politicos are using AI deepfakes to spread lies. The White House is postin
Recent research on voters’ use of AI is forcing campaigns to rethink their digital strategies and reminding election officials of the importance of upkeep regarding information online.
An AI agent doesn't need to want anything to threaten an election — it just needs a goal, some autonomy and the will to keep trying until it succeeds. Read the story on VTDigger here: Where do elections fit in the AI apocalypse? .
The companies may be seeking to score political goals ahead of the U.S. midterm elections
A series of Supreme Court rulings have left the federal government with a narrow path to regulating open-weight AI models. The post The Supreme Court’s Open-Weight AI Regulation Gauntlet appeared first on Just Security .
Tech company leaders' warnings about the dangers of AI demand urgent international legal efforts to regulate artificial intelligence. The post September 12th’s Red Alert: How International Lawyers Should Answer AI Leaders’ Wakeup Call appeared first on Just Security .
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.