Fake news 2.0: personalized, optimized, and even harder to stop
Fake news may have already influenced politics in the US, but it’s going to get a lot worse, warns an AI consultant to the US government.
Sean Gourley, founder and CEO of Primer, a company that uses software to mine data sources and automatically generate reports for US intelligence agencies via In-Q-Tel, the intelligence community’s investment fund, told a conference in San Francisco that the next generation of fake news would be far more sophisticated thanks to AI.
“The automation of the generation of fake news is going to make it very effective,” Gourley told the audience at EmTech Digital, organized by MIT Technology Review.
The warning should cause concern at Facebook. The social network has been embroiled in a scandal after failing to prevent fake news, some of it created by Russian operatives, from reaching millions of people in the months before the 2016 presidential election. More recently the company been hit by the revelation that it let Cambridge Analytica, a company tied to the Trump presidential campaign, mine users’ personal data.
In recent interviews, Facebook’s CEO, Mark Zuckerberg, suggested that the company would use AI to spot fake news. According to Gourley, AI could be used in the service of the opposite goal as well.
Gourley noted that the fake news seen to date has been relatively simple, consisting of crude, hand-crafted stories posted to social media at regular intervals. Technology such as Primer’s could easily be used to generate convincing fake stories automatically, he said, and that could mean fake reports tailored to an individual’s interests and sympathies and carefully tested before being released, to maximize their impact. “I can generate a million stories, see which ones get the most traction, double down on those,” Gourley said.
Gourley added that fake news has so far been fed into social-media platforms like Facebook essentially at random. A more sophisticated understanding of network dynamics, as well as the mechanisms used to judge the popularity of content, could amplify a post’s effect.
“Where you inject information is going to have a massive impact on how it spreads and diffuses,” Gourley said. He went on to suggest that a platform like Facebook may be inherently flawed for sharing news. “All we’ve seen at the moment is primitive, and it’s had a profound impact, and more is coming,” he said.
Gourley did, however, agree that AI would be at least part of the solution. “If machines are going to produce it on one side,” he said, “then you’d better have machines helping you sift through it on the other.”
Correction: An earlier version of this story incorrectly described Primer as a direct contractor of the CIA, rather than of In-Q-Tel.
The inside story of how ChatGPT was built from the people who made it
Exclusive conversations that take us behind the scenes of a cultural phenomenon.
AI is dreaming up drugs that no one has ever seen. Now we’ve got to see if they work.
AI automation throughout the drug development pipeline is opening up the possibility of faster, cheaper pharmaceuticals.
The original startup behind Stable Diffusion has launched a generative AI for video
Runway’s new model, called Gen-1, can change the visual style of existing videos and movies.
GPT-4 is bigger and better than ChatGPT—but OpenAI won’t say why
We got a first look at the much-anticipated big new language model from OpenAI. But this time how it works is even more deeply under wraps.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.