In February, a tragic incident occurred when a Tomahawk cruise missile struck an Iranian elementary school, resulting in the deaths of over 150 individuals, predominantly children. Initially, speculation arose about the origins of the missile, with President Trump suggesting that it might have belonged to another country. “Whether it’s Iran or somebody else, the fact that a Tomahawk—this missile is very generic, it’s sold to other countries, but that’s being investigated right now,” he stated.
Iran released video footage claiming to prove the missile was of U.S. origin. To verify its authenticity, several media outlets consulted Hany Farid, a renowned expert on deepfakes and professor at Dartmouth College. Farid, who also co-founded the cybersecurity firm GetReal Security, explained the challenges of analyzing the footage due to the video’s grainy and unstable quality. Despite these difficulties, the video’s physical plausibility could be assessed using sound speed, light speed, and missile size metrics. Ultimately, this evidence supported the video’s authenticity, despite the complex landscape of digital fabrication.
Farid articulated his concerns regarding artificial intelligence’s ability to create deceptive realities. He emphasized the importance of scientific explanation over reliance on computational outputs in legal settings. “As a scientist, it frustrates me to hear, ‘The computer told me so,’ in judicial contexts,” Farid said. He highlighted the necessity of clear human explanation behind evidence and its implications.
One notable instance involved ICE agents confronting Alex Pretti in Minnesota. Amid chaos on January 24, Farid analyzed enhancement anomalies in footage showing Pretti. The enhanced image misleadingly depicted Pretti holding a gun due to AI misinterpretation of shadows. “The crispness made the scene feel more real, but also resulted in an incorrect depiction,” Farid noted.
The prevalence of fake content on the Internet, often referred to as “AI slop,” raises significant concerns. Farid estimates about half of widely circulated online content is fabricated, drawing attention through sensational AI-generated imagery. Among these contents, deepfake technologies are increasingly weaponized for various purposes—ranging from disinformation campaigns to identity concealment for IT positions.
Farid recounts Fortune 500 companies potentially employing disguised individuals using deepfakes, challenging industry norms of virtual interviews and remote work. He described the exhausting nature of constant suspicion over digital communications. “It’s tiring to always watch your back,” he confessed.
Despite growing difficulties in distinguishing between reality and AI-generated fabrications, Farid remains committed to combating these issues. He returned to Dartmouth and settled in rural Vermont, pursuing peace amid the professional tension. Although isolated, he described the location as serene.
Farid criticized the addictive nature of social media, suggesting profound repercussions across generational learning outcomes. The recent $17 billion settlement by Meta included provisions like a time limit for young users. Farid advocates for a cautious approach toward social media engagement.
Farid anticipates increased autonomy among AI agents, potentially generating fakes independently. He envisions scenarios where AI interacts with humans to achieve deceptive objectives. These developments fuel contemporary debates over AI safety and future implications.
Farid even contemplates AI agents developing profound personalization for future generations, posing existential questions regarding human-AI relationships. “I think there’s a near-future where people will have an AI agent that they grow up with,” he mused.