Over the years, brands have learned (some sooner than others) to listen to what was being said about them in the media, social media, forums and platforms for opinion. Media monitoring allowed measuring media presence; social listening, understand conversations, detect risks and anticipate changes in perception. But the digital ecosystem, which is ever-changing and unpredictable, has incorporated a new intermediary: the assistants of artificial intelligence.
ChatGPT, Gemini, Perplexity, Claude… They are no longer just tools for productivity. Users use them to find out information, compare options, prepare decisions or understand who is who in a sector. In that context, it is not enough to know what people say about a brand. We must also begin to understand what “believes”, summarises and reproduces the AI about itself.
This new layer of listening is known as AI listening: monitoring how AI assistants mention a brand, what tone they use, which sources they cite, and to what extent their responses align with the corporate narrative the organisation wishes to build. To monitoring in the media and on social media, we must now add monitoring within generative AI tools.
From social listening to IA listening
Generally speaking, social listening helps brands understand human conversations, find out what users are saying on social media, how certain topics go viral, what criticisms are being repeated, and which communities influence public perception. This discipline remains useful, but it no longer covers the entire reputation landscape.
The difference is that AI assistants don’t just reflect conversations; they also summarise them, prioritise them and turn them into answers. When a user asks “which are the most innovative companies in this sector”, “which bank has the best reputation”, “which brand is the most sustainable” or “who leads this market”, the AI does not simply display a neutral list of sources. It crafts a response that may include names, comparisons, assessments and omissions.
That is why AI listening is not just about checking whether a brand appears. It involves analysing how it appears. Whether the mention is prominent or incidental. Whether the description is accurate or out of date. Whether the tone is positive, neutral or critical. Whether the sources used are reliable. And, above all, whether the AI’s response aligns with the narrative the company is trying to build.
We are moving from monitoring human conversations to analysing machine-generated narratives. And this shift is strategic because AI assistants are becoming new mediators of public perception.
Why AI listening matters in corporate communications
AI can become a new point of contact between a brand and its audiences. This is why it is important for a communications department. A journalist might use it to prepare for an interview. A job seeker might use it to research a company before applying for a vacancy. An investor might use it to get an initial overview of a sector. A consumer can use it to compare alternatives. A regulator or a stakeholder can use it to put a controversy into context.
In all these cases, the AI-generated response can serve as a first impression. And first impressions are often the ones that matter most.
This forces the teams of communication to ask themselves questions such as: what does ChatGPT say about our company; it mentions us when it asks about our sector; which sources are being used to talk about us?; is the information up to date?; are our key messages recognised? or, is the AI constructing another narrative?
What should be monitored
AI listening begins by systematically asking relevant questions across different assistants and observing the responses. It is not a matter of conducting a one-off test, but rather of repeating queries, recording results and identifying patterns.
The measurement process can begin manually, using a series of questions about the brand, its products, its spokespersons, its competitors and its sector. This method provides an initial qualitative snapshot of the brand’s visibility, helps identify inaccuracies and assesses overall sentiment without the need for significant technical investment.
As the project matures, this measurement can be automated using APIs or by utilising commercial solutions. But the principle remains the same: turning what AI says about a brand into actionable and useful information for communication.
Among the indicators that are most useful are the volume of mentions, the quality of the mention, the sentiment, the consistency of the responses, the alignment narrative and the sources cited.
AI reads newspapers too
One of the major benefits of AI listening is that it helps us understand which sources are ‘training’ the models. And this is where corporate communications has a significant role to play.
Unpaid content carries significant weight: 95% of mentions came from unpaid media, 89% from earned media and 27% from journalistic content. This reinforces an important point for agencies and communications departments: media presence does not only impact human audiences. It can also influence what AI systems read and, consequently, how they respond.
Press releases, although they do not always appear as a direct source, still play a role in the ecosystem: they can generate follow-up coverage, inform journalistic content and build context around a brand. In other words, the traditional work of PR is not disappearing; it is taking on a new dimension.
The risks of not listening
Failing to monitor what AI says about a brand implies accepting a “blind spot” in terms of reputation. The brand may be investing in content, media and positioning, whilst AI assistants continue to provide a description that is incomplete, out of date or poorly aligned with its strategy.
But the risk does not lie solely in invisibility. There may also be errors, inconsistent responses, unwanted associations or an over-representation of old news compared to recent milestones. In contexts of crisis, furthermore, the lack of reliable information can leave room for interpretations that are biased or for content of poor quality.
The materials on deepfakes, disinformation and hybrid threats serve as a reminder that generative AI can also be used to manipulate, spread disinformation or launch more sophisticated attacks against organisations, with reputational damage that is difficult to reverse. Although AI listening does not in itself resolve these risks, it does help to detect early warning signs and understand how a particular narrative circulates or gains traction.
From listening to action
The true value of AI listening becomes apparent when it is linked to the communication strategy. If attendees do not mention the brand, there may be a lack of authority or presence in relevant sources. If they describe it inaccurately, it may be worth reviewing basic corporate content. If they cite unrepresentative sources, it may be necessary to strengthen the brand’s presence in the media, sector reports or well-structured company websites. If the tone does not align with the desired narrative, work will need to be done on messages, spokespersons and content to help balance perceptions.
AI listening is neither a passing fad nor just another layer of reporting. It is a new approach to corporate reputation. Just as brands learned to listen to the media and then to social media, they now need to listen to AI models as well.
