Ai’s shadow reputation: how companies are secretly shaping the narrative

Corporate titans are losing control of their image – not to public scrutiny, but to the algorithms of ChatGPT, Gemini, and Copilot. A new report from FTI Consulting reveals a startling truth: the data fueling these artificial intelligence assistants is overwhelmingly sourced from the companies themselves, granting them an unprecedented influence over how they’re perceived.

The unseen power of annual reports

The analysis, focusing on the IBEX 35, demonstrates a critical shift. The most impactful information driving the AI’s narratives isn’t gleaned from mainstream media outlets, but from corporate pronouncements – annual reports, sustainability statements, and earnings releases. Effectively, these firms are crafting their own digital legacies, often without fully realizing the extent of their control.

This isn’t about a simple data feed; it’s a fundamental power dynamic. Previously, a company’s reputation relied on independent analysis and journalistic investigation. Now, it’s increasingly shaped – and potentially distorted – by a machine learning model trained primarily on its own self-serving statements.

But here’s the chilling caveat: 60% of the reputational risks identified by the AI relate to historical issues – past scandals, legal battles, and controversies that have already been addressed, yet stubbornly persist online. The models dutifully reference these risks, but typically fail to articulate the solutions implemented, leaving a narrative incomplete and invariably detrimental to the company’s brand.

Adding to the complication is the unexpected role of user-generated content. Reddit, Wikipedia, Glassdoor, and LinkedIn are emerging as significant sources of information – often outside the control of corporate communications departments. Reddit surfaces with over a third of ChatGPT's risk assessments for the IBEX 35, while Wikipedia provides the foundational structure for many past-related narratives. LinkedIn and Glassdoor are increasingly dominant in shaping perceptions of company culture and employer reputation.

The technical specifics are telling. ChatGPT, utilizing a broad range of sources including forums and Wikipedia, demonstrates a heightened sensitivity to negative online content. Gemini, leveraging social media and YouTube videos, incorporates a broader, albeit potentially less verified, dataset. And Copilot, embedded within the Microsoft ecosystem, relies primarily on investment-focused media, exhibiting minimal exposure to platforms like Reddit or Wikipedia. The resulting outcomes are stark: a company’s reputation can vary dramatically depending on the AI model used to access it.

The implication is clear: communication teams are potentially overlooking crucial elements of their brand narrative, allowing AI-generated interpretations to dominate. It’s a quiet erosion of control, where the past—even if resolved—continues to haunt a company’s digital footprint.

Ultimately, the report underscores a critical challenge: ensuring transparency and accountability in the age of artificial intelligence. Companies must actively manage their online presence, recognizing that the algorithms shaping public perception are increasingly reliant on their own carefully curated data – a dangerous feedback loop that demands immediate scrutiny.