Largest study of its kind shows AI assistants misrepresent news content 45% of the time – regardless of language or territory

TITLE: AI News Assistants Show Systemic Flaws in Global Study, Threatening Information Trust

Groundbreaking Research Reveals Widespread AI News Distortion

In a comprehensive international investigation coordinated by the European Broadcasting Union and spearheaded by the BBC, artificial intelligence assistants have been found to misrepresent news content 45% of the time across all tested languages, territories, and platforms. This unprecedented study, involving 22 public service media organizations across 18 countries working in 14 languages, exposes systemic issues that transcend geographical and linguistic boundaries.

Groundbreaking Research Reveals Widespread AI News Distortion
Methodology and Scale of the Investigation
The Growing Role of AI in News Consumption
Addressing the Information Integrity Challenge
Audience Trust and Attribution Challenges
Implications for the Future of Information Ecosystems

Methodology and Scale of the Investigation

The research represents the largest study of its kind to date, building on earlier BBC findings from February 2025 that first identified AI’s challenges with news content. Professional journalists from participating organizations evaluated over 3,000 responses from four leading AI platforms: ChatGPT, Copilot, Gemini, and Perplexity. The assessment focused on critical journalistic standards including factual accuracy, proper sourcing, distinguishing opinion from fact, and providing adequate context for news stories., as our earlier report

Jean Philip De Tender, EBU Media Director and Deputy Director General, emphasized the significance of the findings: “This research conclusively shows that these failings are not isolated incidents. They are systemic, cross-border, and multilingual, and we believe this endangers public trust. When people don’t know what to trust, they end up trusting nothing at all, and that can deter democratic participation.”

The Growing Role of AI in News Consumption

AI assistants are rapidly becoming primary information sources for many users, particularly among younger demographics. According to the Reuters Institute’s Digital News Report 2025, 7% of total online news consumers now use AI assistants to access news, with this figure rising to 15% among users under 25. This shift away from traditional search engines toward AI-powered information retrieval makes the accuracy concerns particularly urgent.

Peter Archer, BBC Programme Director for Generative AI, noted the tension between AI’s potential and its current limitations: “We’re excited about AI and how it can help us bring even more value to audiences. But people must be able to trust what they read, watch and see. Despite some improvements, it’s clear that there are still significant issues with these assistants.”, according to emerging trends

Addressing the Information Integrity Challenge

In response to these findings, the research team has developed a News Integrity in AI Assistants Toolkit designed to help address the identified problems. This comprehensive resource focuses on two key questions: what constitutes a good AI assistant response to news queries, and what specific problems need resolution. The toolkit aims to improve both AI response quality and user media literacy.

The EBU and its member organizations are taking additional steps to address these concerns, including:

Advocating for regulatory enforcement of existing laws covering information integrity, digital services, and media pluralism
Pushing for ongoing independent monitoring of AI assistants to keep pace with rapid technological development
Exploring options for continuous research on a rolling basis to track improvements and emerging challenges

Audience Trust and Attribution Challenges

Complementary research published by the BBC reveals a concerning disconnect between public perception and AI performance. The study shows that over a third of UK adults trust AI to produce accurate news summaries, with this figure rising to nearly half among people under 35. This trust is misplaced given the high error rate identified in the main study.

Perhaps more troubling is the attribution problem: when users encounter errors in AI-generated news summaries, they tend to blame both news providers and AI developers, even when mistakes originate from the AI systems themselves. This misattribution could ultimately undermine trust in legitimate news organizations and established news brands.

Implications for the Future of Information Ecosystems

The study’s findings arrive at a critical juncture in AI development and deployment. As these tools become increasingly integrated into daily information-seeking behaviors, the systemic issues identified threaten to erode the foundation of informed public discourse. The cross-border, multilingual nature of the problems suggests that solutions will require international cooperation and standardized approaches to AI training and output validation.

The research underscores the urgent need for collaborative efforts between news organizations, AI developers, regulators, and the public to ensure that the promise of AI-assisted information retrieval doesn’t come at the cost of accuracy and trustworthiness. As Archer emphasized, “We want these tools to succeed and are open to working with AI companies to deliver for audiences and wider society.”