Generative AI models are susceptible to a kind of cyberattack called “data poisoning,” whereby malicious actors intentionally manipulate known source material to change the model’s understanding of an issue. It’s like a high-tech version of giving a school rival a fake exam answer key.
Researchers say that concerns of data poisoning are mostly hypothetical at this point, but showed in a new report how Wikipedia entries could be edited at strategic times to ensure the incorrect information is captured by models scraping the online encyclopedia. It’s an early warning to AI companies and those who depend on it that attackers could soon find creative ways to target the most powerful models and exploit vulnerabilities.
Data poisoning isn’t all bad: Some copyright holders are using a form of data poisoning as a defensive mechanism to prevent AI models from gobbling up their creative works. One program called Nightshade was developed to distort an image when it’s ingested by a large language model.
Mastercard Economic Institute's Outlook 2026 explores the forces redefining global business. Tariffs, technology, and transformation define an adaptive economy for the year ahead. Expect moderate growth amid easing inflation, evolving fiscal policies, and rapid AI adoption, driving productivity. Digital transformation for SMEs and shifts in trade and consumer behavior will shape strategies worldwide. Stay ahead with insights to help navigate complexity and seize emerging opportunities. Learn more here.
Think you know what's going on around the world? Here's your chance to prove it.
A Democrat won Miami’s mayoral race for the first time in nearly 30 years. The Republican defeat will ring some alarms for the party – and their support among Latino voters.
50: Mexico’s President Claudia Sheinbaum is taking a page out of US President Donald Trump’s book, implementing up to a 50% tariff on more than 1,400 products in a bid to boost domestic production.