Algorithms have long been able to produce basic news stories from press releases or sets of financial data; that’s not much of a threat to most humans in the news business. Now, however, artificial intelligence (AI) has taken a step further. It’s learned to perform a tougher task — to produce convincing-looking fake news. Stringing together a few formulaic passages from a set of numbers is a mechanical job. Inventing a fake news story on a random subject requires imagination; not every human is up to it. The San Francisco-based non-profit OpenAI, founded by Tesla CEO Elon Musk and Y Combinator president Sam Altman, has produced a so-called language model that can do it. The quality of the output is somewhat uneven, but the best examples resemble human writing to a frightening degree. On the surface, GPT-2, as the model is called, works somewhat like a popular game one can play with the less advanced version of AI on any smartphone, accepting its word suggestions one after another to...

BL Premium

This article is reserved for our subscribers.

A subscription helps you enjoy the best of our business content every day along with benefits such as exclusive Financial Times articles, ProfileData financial data, and digital access to the Sunday Times and Times Select.

Already subscribed? Simply sign in below.

Questions or problems? Email or call 0860 52 52 00. Got a subscription voucher? Redeem it now