Synthetic reputation: managing the story AI tells about your organisation

What is synthetic reputation?

For decades, reputation followed a predictable chain: an organisation acted, journalists reported, and audiences formed opinions. Today, communication professionals face a new reality. Stakeholders are increasingly learning about your brand directly through AI models without traditional media intervention.

Synthetic reputation is the version of your organisation living inside ChatGPT, Gemini, Claude, or other Large Language Models (LLMs); an algorithmic blend of your website, past media coverage, forums, and transcripts. Just because this impression is generated by AI rather than a human mediator does not mean it is any less critical for comms teams to actively manage. 

AI plays by different rules

Medianet's Opening the Black Box Phase 2 research, based on 1,357 AI responses and 11,127 hand-coded citations, shows just how differently reputation works inside the machine.

• The sources: Owned content such as press releases, blog pages, websites, and media centres dominated LLM responses with 67% of citations, while major Australian newspapers being paywalled in Australia made up just 4%.
  
  

• Share of voice: Or how we like to call it, Share of Model. An organisation with a solid share of the conversation on traditional media might not hold a similar share on LLMs. In our study, insurer Youi held 26% SOV in AI, compared with 7% in traditional media, while RACV, ranked first in traditional media presence at 88%, but that dropped to 42% in LLMs.
  
  

• Sentiment: LLMs’ preference for owned structured content has an impact on the sentiment of the responses. For financial services brands, for example, sentiment played out differently in AI mentions as opposed to traditional media. It should not come as a surprise that when the source is corporate content, sentiment skews positive and negative issues are often neutralised.
  
  

• The ‘new’ news cycle: We are used to a short news cycle that usually fades within 4 weeks. That is not the case with AI, which remembers everything and often surfaces information that is years old. Yes – those issues that are done and dusted in the media are still brought back to the surface in AI responses.  

You can't take an AI out for coffee

Not long ago, if a story was heading the wrong way, you'd call the journalist or take them for coffee and talk it through. With AI, there's no one to call. You can't pitch to an LLM; you can only shape what it draws from.

That matters, because the AI answer is often the first impression. Models favour what's plentiful over what's accurate, so outdated narratives can stick, and recent changes can take months to register.

So what replaces the coffee? Auditing AI answers the way you monitor press coverage, tracking sentiment, sources and share of voice. Investing in open, crawlable owned content, including video transcripts. Add an AI check to crisis playbooks so issues are tracked for months, not weeks. And keep investing in journalists, whose work still shapes what AI draws from.

None of this replaces traditional reputation management. But treating AI outputs as a governance and measurement priority, not a technological curiosity, will be one of the defining comms skills of the next few years.

Learn more about how to shape your organisation’s synthetic reputation at IABC’s APAC Fusion event.