When AI becomes your reputation analyst

This is the latest in a series of articles to celebrate PRmoment's PR Masterclass: AI in PR, where we talk to experts about how AI is changing the PR sector. If you want to understand more about the how AI is changing the output and make up of PR teams, often for the better, come along to PR Masterclass: AI in PR.

When AI builds a picture of your organisation, it doesn't just read what you choose to say. It cross-checks it against everything else.

Cloudflare announced that AI agents have surpassed human traffic for the first time in history. By their count, agents are now 57.4% of internet traffic to 42.6% humans. The direction is clear. AI acting on behalf of humans will do most of the web searching in the future.

This has important implications for how reputation is managed. Ask an AI today about a company and it may return something closer to a verdict than a summary. It will analyse multiple sources and cross-reference what you say, what you appear to be doing and what others say about you. From this, it can surface contradictions. Someone using AI can form an opinion about you before they have interacted with you at all.

How AI assembles a reputation

Most of the emerging advice on AI and reputation focuses on visibility and how to get cited in the answer. Being cited is useful. Being cited in a way that damages reputation is a problem.

When someone queries AI about an organisation, it draws on three source types. The first is the declared narrative, what the organisation says about itself via its website, company pages and leadership bios. The second is observed behaviour, what it appears to be doing, drawn from hiring patterns, job listings, product pages and public statements from its people. The third is external validation, what others say, across coverage, reviews and social media. Where these sources contradict each other, AI can surface the inconsistency.

A company can put culture at the centre of its comms, but if its Glassdoor ratings have decreased, its HR team is advertising the same roles on rotation and it recently announced a wave of redundancies, AI can highlight the inconsistencies.

Why communications teams need to own this

For communications teams, the implication is structural. Reputation in AI is assembled from signals that cut across PR, HR, investor relations, legal and product. No single function controls all of them, which is why AI reputation can end up unmanaged.

Communications is the natural owner, not because it controls every signal, but because it is the only function responsible for how the whole picture holds together. All the usual comms metrics still matter, but AI alignment is a new measure that should sit alongside them.

An important caveat is that not all AI results are the same and not all are accurate. Different models can produce different answers to the same question, and the same model can produce different answers depending on how the question is framed.

AI can also hallucinate. It may attribute a quote, a source or a contradiction that does not exist. The only defence is making the accurate signal strong enough that it has better evidence to draw on.

How to test what AI says about you

These are reasons to test AI reputation more rigorously. Communications teams should not only ask the questions they hope stakeholders will ask. They should ask the questions a sceptical journalist, activist investor or competitor might ask as well. The objective is not to find the most flattering result, but to identify the inconsistencies that others are most likely to discover.

It starts with a search, run while logged out and across more than one model. Go beyond the questions you hope stakeholders will ask. Try:

• What contradictions do you see in this company's public record?

• What would a sceptical investor question?

• What would a critical journalist investigate further?

• Where do employee, customer and company narratives diverge?

Which sources are shaping the answers? Where does the picture diverge from the story the organisation tells about itself? The divergences are the work.

If AI produced an assessment of your organisation today, drawing on everything it could find, would it be the assessment you would want it to write? If not, the place to start is not the message. It is the gap between what you say, what you do and what others say about you.