News report: integrity in AI assistants

For the report is one of the largest cross-market evaluations of its kind. Working with the EBU, 22 public service media organizations in 18 countries – working in 14 languages – assessed how ChatGPT, Copilot, Gemini, and Perplexity answer questions about news and current affairs.

The research built on an earlier BBC study, which exposed inaccuracies and errors in AI assistants’ output. This new study explored whether this had improved and if the issues previously identified were isolated or systemic. Alarmingly, it found that AI routinely misrepresents news content, no matter which language, territory, or AI platform is tested. 

The work involved professional journalists from participating PSM evaluating more than 3,000 AI responses against key criteria, including accuracy, sourcing, distinguishing opinion from fact, and providing context.

Some key findings: 

• Almost half of all AI answers had at least one significant issue.
• A third of responses showed serious sourcing problems.
• A fifth contained major accuracy issues, such as hallucinated and/or outdated information.

Find more information in the source description. 

Source: https://www.ebu.ch/research/open/report/news-integrity-in-ai-assistants