AI Pulse · AI Trends Pulse

Two independent audits published on the same day have raised a basic but serious question about AI search: can you trust the citation, or only the answer’s confidence?

The audits found examples where Perplexity cited manufactured SEO-farm pages and cases where the cited source did not actually support the number or claim presented. These were separate investigations using different methods, which makes the overlap more meaningful. The discussion gained attention in the developer community, with the two Hacker News posts receiving 227 and 119 points at the time of reporting, as covered in the original discussion.

For a business owner, the practical lesson isn’t to ban AI search. It’s to change where you place trust. AI search can be useful for finding leads, summarising a topic, or giving your team a starting point. It should not be the final authority for pricing decisions, market sizing, legal guidance, competitor claims, or anything going into a board pack. Ask your team to open the cited links, check the original data, and treat a polished answer as a draft until it’s verified.

This matters beyond Perplexity. Every major AI lab is competing in AI search, and the pressure to produce quick, complete answers can expose weak source checking. This is the kind of thing we build into an AI command centre, with clear review steps for high-stakes outputs rather than blind trust in whatever the model returns.