Study Reveals AI Assistants Distort Up to Half of News Content

AI assistants found to distort news content, affecting public trust.

Research conducted by the BBC in collaboration with the European Broadcasting Union (EBU) has revealed that AI assistants regularly distort news content.

According to the report, AI-based assistants are replacing search engines for millions of people. Seven percent of online consumers use AI for news, with the figure reaching 15% among those under 25.

The extensive monitoring involved 22 public media organizations from 18 countries. Professional journalists evaluated over 3,000 responses from ChatGPT, Copilot, Gemini, and Perplexity based on key criteria, including accuracy, source reliability, separation of opinion from fact, and provision of context.

The study uncovered numerous systemic shortcomings in the four leading AI tools:

• 45% of AI responses contained at least one significant issue;
• in 31% of cases, assistants cited non-existent or misleading sources;
• 20% of responses included inaccurate data, unrealistic details, and outdated information.

Gemini performed the worst in tests, with 76% of its responses containing significant problems, more than doubling the rates of its competitors. Experts attributed this to its low effectiveness in finding providers.

Comparing the study’s results with data from the first phase earlier this year showed that the error rate of AI agents has decreased but remains “still high.”

“The study convincingly shows that such shortcomings are not isolated incidents. They are systemic, cross-border, and multilingual, and we believe this undermines public trust,” said EBU Deputy Director General Jean Philip de Tender.

According to a BBC survey, more than a third of British adults trust AI-generated news summaries. Among respondents under 35, this figure reaches half.

The research team believes these findings only underscore the importance of the identified issues.

Experts have presented a set of tools to improve AI assistants’ responses to news topics and enhance users’ media literacy.

Earlier, scientists identified signs of degradation in AI models due to the use of social media content.