Why “knowledge integrity” matters in the age of AI assistants

The University of Cambridge ThinkLab is partnering with the Wikimedia Foundation on a research project to replicate and expand on the BBC’s investigation into how AI assistants represent trusted information sources. More details on this latest collaboration can be found here on our ThinkLab website: https://www.thinklab.strategic-partnerships.admin.cam.ac.uk/projects/project/new-wikimedia-bbc-financial-times-and-the-future-of-knowledge-integrity/

I feel like this work is necessary and urgent there is no doubt AI is shaping how we seek information, how we encounter that information, and should seriously be challenging us to think carefully about how much we trust where our information is coming from. But if the answer is detached from its provenance, we lose the trail that makes information trustworthy: who said it, where it came from, what evidence supports it, and what uncertainty remains. In other words: context, context, context. As Patrick Gildersleve puts it in a recent LSE blog post: “The advent of LLMs marks a shift towards personalised summaries divorced or obfuscated from their provenance.”

That knowledge integrity question isn’t academic nitpicking — this is a very practical trust issue. Large-scale evaluations led by the BBC and the EBU have found that AI assistants can misrepresent news content at scale, including problems with accuracy and sourcing. When an AI response confidently cites a “trusted” outlet or a reference platform like Wikipedia, the stakes get even higher: attribution can be misleading, citations can be thin or absent, and nuance can be flattened into a single authoritative-sounding paragraph.

At the same time, Wikipedia remains a uniquely transparent knowledge commons: built through debate, citations, and visible editorial history. The Wikimedia Foundation argues that “Wikipedia’s role as the backbone of all knowledge on the internet has never been more important.” And as Jimmy Wales notes, AI can also be used to strengthen that ecosystem: “what’s really exciting about AI is the potential that we might find some ways to use AI to support our community in Wikipedia.”

Further reading

• LSE Impact Blog: https://blogs.lse.ac.uk/impactofsocialsciences/2025/11/17/grokipedia-falls-flat-but-ai-is-already-rewriting-wikipedias-future/

• BBC Science Focus: https://www.sciencefocus.com/news/wikipedia-ai

• Wikimedia Foundation: https://wikimediafoundation.org/news/2025/11/10/in-the-ai-era-wikipedia-has-never-been-more-valuable/

More details on this latest collaboration can be found here.