Weaponized Deepfakes Threaten Society with Political Manipulation and Harmful Content

Deepfake technology has evolved from theoretical warnings to widespread real-world threats, with AI-generated videos and images now easily created using accessible generative models. These weaponized deepfakes range from sexually explicit material to political propaganda, with a 2023 study revealing 98% were pornographic and 99% depicted women. Specific examples include Elon Musk's xAI chatbot Grok, whose image editing function produced millions of sexualized images, 81% depicting women according to one report. Political manipulation has become rampant, with the Trump administration regularly sharing AI-generated content, including an altered image of Minneapolis civil rights lawyer Nekima Armstrong that darkened her skin and changed her expression. In January, Texas attorney general Ken Paxton shared a fabricated video showing his opponent Senator John Cornyn dancing with Representative Jasmine Crockett. Proposed solutions like technical safeguards, user behavior changes, and legislation face significant limitations, as open-source models bypass restrictions and enforcement remains inconsistent. The problem may intensify during upcoming U.S. midterm elections, with weakened federal agencies and diminished fact-checking organizations creating vulnerabilities. These developments threaten critical thinking skills and institutional trust while disproportionately impacting women and marginalized groups.