Concerns about misinformation and the rapid rise of artificial intelligence took center stage at Raynham Hall Museum, where technology expert Scott Lowenstein delivered a presentation titled “Under Cover: 250 Years of Spying in America: Misinformation in the Age of AI.”
Lowenstein, director of product for AI at The New York Times and a former leader of the News Provenance Project, traced the evolution of misinformation from historical examples to modern digital challenges at the presentation on Monday, emphasizing both the novelty and continuity of the issue.
He described how his interest in misinformation begin during his work as a pollster for political campaigns, then deepened during graduate school in the aftermath of the 2016 election, when public discourse increasingly focused on “politics, misinformation, fake news.”
Lowenstein later worked at Twitter on efforts to improve online discourse before joining The New York Times to address questions of trust in media.
“Because obviously the New York Times is a company,” he said, “and that’s built on trust and the relationship that you all have with us.”
A central focus of the talk was how artificial intelligence has accelerated the spread and sophistication of false or misleading content. Lowenstein highlighted a widely circulated AI generated image of Pope Francis from 2023 wearing a fashionable coat as an early example of the technology’s viral potential.
He demonstrated how similar images can now be created in seconds using widely available tools, warning that improvements in quality are making such content harder to detect.
“The possibilities of synthetic media,” Lowenstein explained, “and the risk of someone creating something claiming it’s from us or us being fooled by a false image, has just grown exponentially.”
Despite the technological leap, he argued that misinformation itself is not new, drawing parallels to earlier eras. He cited examples ranging from altered historical photographs to misleading newspaper headlines and propaganda.
“Anytime there’s a new sort of mass communication technology, people take advantage of it because it’s amazing to spread new ideas,” he said. “And then as soon as something takes off, someone is figuring out how to use it to their advantage.”
Lowenstein also distinguished between misinformation and disinformation, noting that intent plays a key role. Whereas misinformation is falsified or misleading information spread naturally and without negative intent, disinformation is created is disseminated with clear negative purpose.
He pointed to subtle manipulations, such as slowed video altering perception, as examples of how even minor changes can significantly impact public understanding.
A recurring theme of the presentation was what Lowenstein called the “liar’s dividend,” the idea that widespread skepticism can undermine trust entirely.
“If you can just say, Oh, that was fake, or Oh, that was AI generated… then what is true anymore?” he asked.
To address these challenges, Lowenstein outlined several approaches, including detection tools, improved media literacy and the concept of “media provenance,” which tracks the origin and history of digital content.
However, he acknowledged that no single solution is sufficient and that each approach has limitations, particularly as technology continues to evolve.
The presentation concluded with a discussion of how institutions like The New York Times are trying to adapt to AI.
“We treat it as a tool,” Lowenstein said, “not a replacement for people.”
