How to stop AI chatbots going rogue
Large language models like Grok, ChatGPT and Gemini don't know what's true or harmful.
At Full Fact, we believe in asking tough questions about how these systems are built and deployed.
Full Fact fights bad information
Bad information ruins lives. It promotes hate, damages people’s health, and hurts democracy. You deserve better.
Recent fact checks
Google Lens AI overviews have been giving users false and misleading information about certain images shared widely on social media.
A Full Fact investigation found the AI overviews for at least ten clips failed to identify certain images as inauthentic or shared misleading information about them.
Articles published by the Guardian and the Telegraph claimed that nine in 10 nurses rejected the NHS staff pay award for this year.
Ninety one percent of Royal College of Nursing members who took part in a vote said the pay award was not enough—but only 56% of those who were consulted actually voted.
A government minister claimed “we’ve already brought the number of asylum hotels down from 400 at the peak of the previous government to 200 now”.
Almost all of this decrease took place before Labour took office. In its first year, Labour reduced the number of hotels housing asylum seekers by a net of three.
Can you help us do more?
Every donation will help us stop the spread of bad information.