6 things comms leaders need to know about reputation management in the age of AI

By Kate Neilson, IABC Board Director

More than half of all web traffic today is machine-to-machine, and search behaviour is shifting fast – from typing queries into Google to simply asking an AI model for the answer.

For communications professionals, this shift changes almost everything about how reputation is built, monitored and protected online.

At a recent IABC NSW webinar, a panel of experts – featuring Amrita Sidhu, Managing Director of MediaNet; David Hickey, Executive Director APAC for Meltwater; and Kate Hamilton, Vice President Advisory APAC for Reptrak – walked through what this new landscape looks like in practice, moving from success stories to crisis playbooks to the mechanics of how different AI models source their answers.

Here are some of their key takeaways for comms professionals:

1. Audit your owned content, it matters more than you think

‍Sidhu's research across Australian retail banking, insurance and government sectors found that owned content is doing far more heavy lifting than communicators may realise.

‍"Sixty-seven per cent of the sources were actually owned content," she said - meaning websites, PR material, corporate newsrooms and informational archives — with earned media accounting for a further 16 per cent.

‍Part of the reason owned channels carry so much weight is that they're the one part of the ecosystem an organisation fully controls. Earned media citations depend on factors well outside a comms team's hands - whether a publisher's content is even accessible to AI crawlers, whether a licensing deal with that AI provider exists yet, and how a journalist chose to frame a story.

‍Owned content has none of those obstacles: it's open to being crawled and the organisation decides exactly what it says, how it's structured and when it's updated.

‍ That makes it the most direct lever communicators have for shaping how they show up in AI-generated answers.

"Owned media is incredibly powerful for communication and corporate affairs teams, as a way of trying to start to understand how they can influence how they're showing up in AI," said Sidhu.

‍Don't assume global AI-visibility research applies here.

"Australia is a very heavily copyright-regulated market as compared to some of the other markets," Sidhu explained.

‍Much local media sitting behind paywalls and active AI crawler blockers. That makes generative engine optimisation "slightly different in this market" - and makes owned channels even more important in the meantime.

2. Remember that AI never forgets

‍Both Sidhu and Hamilton flagged the same risk: large language models have an unusually long memory.

"GenAI has this really long memory span," said Sidhu. "It goes back and it references content that you've produced a few years ago... our content now could potentially be remembered by AI forever."

‍Hamilton warned this can resurface old context without important nuance.

‍"You see that, 'Oh, great, it's [an article] from ABC, but then you follow through and see it's an article from 2021."

‍She suggests regularly reviewing and refreshing older website content rather than letting it sit indefinitely.

3. Treat AI as a stakeholder, not just a channel

‍Hamilton argued comms teams need a mindset shift. AI isn't merely relaying facts - "it's actually giving you the answer, and giving you opinions on what you should do... whether it's what bank you should go with, what health insurer you should pick."

‍That makes it, in her words, "its own stakeholder" with real influence over decisions, not a passive touchpoint to be monitored.

‍Hickey pointed to a live example of that influence playing out in financial services, where Commonwealth Bank has been "leveraging AI to be able to speed up their ability to solve customer problems, to be able to add more context via conversational agents to understand the customer better, to be able to make the experience better."

‍In other words, the AI is actively shaping the customer's experience of, and opinion of, the brand in real time, much like a human stakeholder would.

‍This influence will only become more embedded, and less visible, over time, according to Hickey.

‍"Mid-term, we think that people won't worry too much about whether AI is in the loop," he said. "They're just going to worry about whether or not their customer service has been improved, and whether or not the customer service that they can get from Bank A is going to be better than it would be from Bank B, C, or D, or any other industry for that matter."

‍Customers may never mention "the AI" at all, but their opinion of the brand will be shaped by it regardless. 

‍"AI isn't just a technology conversation, it's also a people conversation," said Hickey. ‍

4. Don't chase one dominant platform

There's no single AI platform to optimise for, and every model behaves differently.

‍Hamilton laid out the basic problem - different platforms are trained on entirely different corners of the internet.

‍"Gemini, for example, pulls from YouTube, whereas ChatGPT doesn't," she noted, adding that the same organisation conveyed completely differently depending on what's being asked: a question about "Should I take an interview at this particular organisation?" draws on different sources than "Should I buy this product?", for example.

‍Hickey said the fragmentation of the landscape also influences outcomes.

‍"Claude doesn't [pull from] Reddit, but Gemini and ChatGPT do. Grok has X, and no one else will have X."

‍His view is that this patchwork of data sources should directly inform strategy - not just for AI visibility, but across the board.

‍"There's a lot you can learn about how LLMs are behaving based on where they're actually getting their data from," he said. " I think that that will really help to inform your strategy, and even help you understand whether your strategy in LLMs might align with your strategy in social and earned."

‍Sidhu framed the whole challenge as a "black box", as responses depend on the questions you put, the industry, the prompts, where the content's being sourced from. That opacity is exactly why she'd caution against picking a favourite.

"I just don't know that we have enough information at this stage to be like, "I'm doubling down on Claude" - not unless an organisation has already done deep AI visibility research into its own citation sources and content strategy," she said.

‍"I would say this is the most important time to learn, experiment, share as an industry, understand what's going on."

‍Hickey agreed, stating that the starting point should always be the audience, not the platform.

‍"You've got to start by understanding who your audience is… and where does that audience live," he said. "Is that audience on Claude, or is it on ChatGPT? If it is in only one place - like, say, for instance, a business audience, and they live on Copilot - then maybe that might help you index in that direction." ‍ ‍

5. Close the loop publicly when things go wrong

‍When facing a reputation or brand crisis, it's far less likely that going quiet and hoping the story fades away is going to work in an AI-indexed world, said Hamilton.

‍Because LLMs are constantly crawling and indexing content - and, as noted above, hang onto it indefinitely - a story doesn't simply drop out of circulation the way it might have in a traditional news cycle; the original coverage stays retrievable and citable long after public attention has moved on.

‍"Traditionally… we might have said, "Let's just be quiet for a bit and see if it blows over," said Hamilton, "Whereas that's almost the opposite of what you need to do when we're talking about this AI world."

‍Organisations that clearly explain how they resolved an issue and what they learned are rewarded by LLMs as "a positive contributor" to reputation.

6. Get ahead of misinformation using your own website

A false narrative can go global before the truth gets its boots on, said Hickey.

He pointed to a real client facing this challenge. It was a global company operating in Australia, fighting off persistent false claims about its projects. Rather than chasing every social media post, the team is focusing on using their website and their own real estate to get the correct version out there and keep it visible, he said.

‍Practically, that starts with tracking where the false narratives are actually cropping up in LLM answers, then making sure the site is structured well enough that AI models pick up the correct version instead.

‍"It's probably the best example I've seen of someone being really proactive with trying to correct some of these issues," Hickey said.

‍He also floated a more systematic version of the same idea. Train a listening tool on your own corporate truths, then set up alerts for whenever a social post or LLM answer contradicts them, so you know how far a mistruth has actually spread before deciding how hard to push back. Worth keeping an eye on too, he added, is Reddit's on-again, off-again relationship with the big models.

‍"It's been a bit up and down in terms of whether LLMs are drawing from Reddit... but that's where crises or issues can sort of start simmering." ‍ ‍

The bottom line

If there's one thread tying the whole conversation together, it's that reputation has stopped being something a comms team can quietly manage from behind the scenes.

‍"Reputation and building... that perception of your organisation is now multiplayer" - no longer a one-way broadcast, but an ongoing negotiation between what an organisation puts out, what real people say, and what AI models decide to say on its behalf," says Hamilton.

‍That shift is unsettling, but the panel agreed it's also an opening for communicators to highlight the expertise and value they bring.

‍"I think there is a real strength and a real power here for comms teams," said Sidhu.

‍And with the field still this new, there's genuine advantage on offer for those willing to get in early. As Hickey put it, "we're still pretty early days" - which means the organisations that start auditing, experimenting and building real AI visibility now won't just be playing catch-up later. They'll be setting the standard everyone else has to meet.

If you're an IABC member and you'd like to watch an on-demand version of this webinar, reach out to contact@iabcnsw.com and we will send you the link.