The Ultimate PR Stack: Best Tools for PR Planning, Social Listening, and Influencers PRmoment Health and Wellness Comms Awards 2025 PR Masterclass: AI in PR PRmoment Leaders

Leveraging LLMs for Smarter Earned Media

Credit: Andrea de santis via unsplash

The role of earned media in reputation has received a boost in its relevance, thanks to the way Large Language Models (LLMs) like ChatGPT, Gemini and Perplexity operate.

A recent report 'Coverage to Capital: Reputation & AI ' by marketing and communications agency,  'Hard Numbers', says, "Editorial media remains the most influential source across all four all four brand traits (Quality, Trust, Innovation, and Value.)  we analysed, making up 61% of the total responses generated by LLMs. 

The report also highlights that earned media was especially influential when discussing Trust (65%) and Value (72%).

Sign up for a virtual ticket for our AI in PR masterclass. 10 expert sessions. Recording available.

How Gen AI amplifies the value of earned media

Ishu Bansal, co-founder. Sarvodaya Ventures agrees with the report by Hard Numbers. 

He explains, "Gen AI amplifies the value of earned media by making high-authority, organic mentions more likely to appear in AI-generated answers. Brands can leverage this by focusing on credibility: creating authoritative content, securing quality backlinks, and appearing in trusted publications. PR teams should also optimise for inclusion in structured knowledge graphs (e.g., Wikipedia, Wikidata), as these are key sources for many AI models."

Bansal adds, "Content today must also be machine-readable, context-rich, and structured to appeal to generative AI outputs. PR teams are now optimising press releases and thought leadership for keywords, semantic clarity, and data enrichment. This includes formatting content with clear context, using structured Q&A formats, and ensuring factual accuracy—all to improve visibility when LLMs surface content in responses."

Prepping LLMs with GEO, why does it matter for brands?

Sushobhan Chowdhury, founder of AHA, explains why consumers are increasingly using LLMs for search. 

He points out that, "GEO is democratising the overall approach of how I'm getting answers.  If I ask, 'Which is the best sunscreen for my child?' What will Google do?

Google will say, 'Hey, we've indexed on certain parameters and we believe that this particular site answers that best. Which means Google is creating these rules to say if it answers A, B, C, D, E, whatever it is, then we know that it is answering the best. However, I will not limit the user to just looking at that. I'll give the limit the user another five options to say you figure out what you think is right.

The fundamental change now is in the fact that it is now convergence. You can go directly to the answer via LLMs and GEOs without trawling through several pages.

However, you can also say, 'Don't give me one singular answer; give me a comparison of the top 5 choices', then GEO will give you that comparison of 5.

This change reshapes the content opportunity for brands, both in terms of earned, owned, but also how the product pages are displayed and enriched with answers. 

Sign up for a virtual ticket for our AI in PR masterclass. 10 expert sessions. Recording available.

Role of Owned Media in Generalised Engine Optimisation (GEO)

One interesting outcome of LLMs has been how owned media is also resonating with LLMs and GEO. 

While earned media matters for LLMs, Bansal explains that, "AI powers conversational interfaces, intelligent search, dynamic content personalisation, and smart assistants. UI/UX design is evolving to support these interactions through natural language inputs, multimodal interfaces (text, voice, images), and adaptive layouts.

Transparency is also crucial: users now expect trust signals like citations, human oversight, and clear AI disclosures in the interface."

Owned media and the naivety of AI LLMs - Hard Numbers Report

The 'Coverage to Capital: Reputation & AI '  report says owned media—brand-generated content such as blogs, press releases, and company websites—shapes LLM responses, especially around Innovation (66%) and Quality (55%). "

The study also revealed a striking extent to which "LLMs accept owned media when determining whether a company is trustworthy. When asked, “Is this company trustworthy?” LLMs often respond with, “Well, they say they are, so yes.” It’s as if the AI is accepting the company’s word without question. This is a surprising dynamic in a world where we expect AI to offer more impartial evaluations, yet LLMs seem to take brand-generated content at face value."

The Hard Numbers study warns, "While owned media had much less influence on Trust (4%) and Value (4%), the fact that it dominates responses related to innovation and quality suggests that LLMs are still heavily influenced by what companies say about themselves. This could allow brands to shape their narrative without much external scrutiny, highlighting a potential blind spot in AI’s ability to challenge self-promotional content."

It also raises issues about copyright and user data.

Mainstream media and licensing agreements with LLMS

International media publications like the 'Financial Times' have licensing agreements with LLMs like ChatGPT, which ensures that their content is being used with compensation.

Germaine Pereira,  partner with law firm Solomon & Co. and a specialist in media and entertainment as well as intellectual property explains how such an agreement works, "Such agreements display, attribute and maintain the publisher’s credibility which in turn benefit publishers as well as the licensee in many ways including (a) stipulating the use of content, (b) protection of content against unauthorized use, copying, or infringement or plagiarism, (c) minimizing copyright infringement disputes, (d) generate revenue through license fees and royalty, (e) helps build brand awareness and loyalty and (f)ensure that both parties comply with relevant laws and regulations."

Indian Media and ChatGPT

In India, large media houses such as ANI and NDTV have filed a case against Microsoft-owned OpenAI for copyright violations. In its response in February this year, OpenAI, as per Reuters, stated that it did not need permission to use publicly available content to train its models. The opinion on whether publicly available equals consent is a hotly debated topic.

Pereira's view is that, "Considering the number of cases claiming use of copyrighted content without authorisation, it is advisable that Indian media houses enter into licensing agreement to license content owned by them to licensees interested in republishing such content in order to specify ownership rights, copyrights and potential liability issues for misuse of original material."

Brands and Licensing Agreements with LLMs

With GEO beginning to emerge as a strong competitor to SEO, should brands also look at licensing agreements with generative AI platforms,  and what does the policy or law say about that?

Pereira explains that, "While a specific legal mandate for licensing agreements may not exist, the complexities of AI and intellectual property raise concerns about user consent, data protection, and privacy. Brands can enter into licensing agreements to protect their interests and ensure compliance with evolving regulations, including the Law of Contract, Copyright and the Digital Personal Data Protection Act.

SEO vs AEO vs GEO

How is the same piece of content prepped for GEO, AEO and SEO? Ishu Bansal explains.

Take a product launch announcement:

  • SEO: The content includes keyword-optimised headers, metadata, and backlinks, targeting specific search intents.

  • AEO (Answer Engine Optimisation): The same content is reshaped into FAQs, snippets, and structured formats (like lists or tables) to answer questions on platforms like Google’s featured snippets or Bing Chat.

  • GEO: The content is further adapted to be LLM-friendly: clearly attributed facts, source citations, and concise, natural-language answers to likely user prompts—so it can be surfaced by AI models like ChatGPT or Perplexity.

If you enjoyed this article, you can subscribe for free to our weekly event and subscriber alerts.

We have four email alerts in total - covering ESG, PR news, events and awards. Enter your email address below to find out more: