The AI News Error Epidemic: What You Need to Know

The rise of AI-generated news has brought convenience and speed to our lives, but at what cost?

TL;DR
• 45% of AI-generated news responses contain serious errors.
• 81% of AI-generated news responses have at least one issue.
• Users can protect themselves by asking for sources, time-stamping queries, and cross-checking.

Summary
AI-generated news responses are increasingly being used, but a recent study by the European Broadcasting Union (EBU) raises concerns about their accuracy. The study found that 45% of AI-generated news responses contain serious errors, and 81% have at least one issue, such as outdated information, misleading phrasing, or missing sources.

The study tested over 3,000 AI-generated responses across 14 languages, including popular AI assistants like ChatGPT, Google Gemini, Microsoft Copilot, Claude, and Perplexity. Key findings include:
• 45% of responses had at least one significant error
• 81% had some form of issue
• 31% were flagged for sourcing problems
• 20% contained major factual inaccuracies

The use of AI assistants as a primary source of news is becoming more common, especially among younger users. However, many of these assistants do not clearly surface sources or distinguish fact from opinion, creating a false sense of confidence. To mitigate these risks, users can take steps such as asking for sources upfront, time-stamping queries, cross-checking information, and not treating chatbots as authorities.

The EBU report warns that the widespread consumption of flawed or biased summaries could distort public discourse and undermine trusted news outlets. To address these concerns, there is a need for greater transparency, stronger sourcing systems, and smarter user behavior. Until AI chatbots can consistently cite, clarify, and update their sources in real-time, users should approach their responses with caution.

As AI continues to shape our news consumption, it's essential to remain vigilant and critically evaluate the information we receive.