In short
- What AI in PR is — in one paragraph: Artificial intelligence in public relations is a set of tools that take over repetitive and computational work: they read thousands of publications, organise mentions, draft texts,…
- AI in PR in the US: data from 2025–2026: The US market is the reference point, because that is where the largest PR platforms, industry bodies and regulators were first to describe the effects of…
- 12 uses of artificial intelligence in public relations: The table below shows what the machine actually does in each use case, and what must stay with a person.
- Brand visibility in AI: a new job for PR: More and more buying decisions start with a question asked to ChatGPT, Gemini, Claude or Perplexity rather than to a search engine.
AI in PR (artificial intelligence in public relations) means using language models and data analytics to speed up media monitoring, sentiment analysis, content preparation and crisis detection — while decisions, relationships and accountability stay with people. In 2026, AI in PR has a second, more important dimension: earned media coverage is what determines what ChatGPT, Gemini, Claude and Perplexity say about your brand.
This guide brings together data from the US market (the most mature in adopting AI in communications), the ethical and legal rules that apply in the US and the European Union, and a practical 90-day implementation plan. Every figure comes with its source and the date of the study.
What AI in PR is — in one paragraph
Artificial intelligence in public relations is a set of tools that take over repetitive and computational work: they read thousands of publications, organise mentions, draft texts, suggest which journalist to approach with a given story, and warn you when negative comments start to grow. What they do not take over is what actually decides the outcome in PR: judging the situation, relationships with people, accountability for every word and a sense of timing. In short: AI will not replace people in PR. It will replace guesswork.
AI in PR in the US: data from 2025–2026
The US market is the reference point, because that is where the largest PR platforms, industry bodies and regulators were first to describe the effects of AI. The figures below come from surveys — they show what practitioners report, not a measurement of their work.
| Indicator | Value | Source and date |
|---|---|---|
| PR professionals using generative AI | 76% (no significant change year on year) | Muck Rack, State of AI in PR 2026, 564 respondents, December 2025 |
| Working at organisations with AI usage policies | 51% (21% in 2024) | Muck Rack, State of AI in PR 2026 |
| Using at least one paid AI tool | 75% (57% a year earlier) | Muck Rack, State of AI in PR 2026 |
| Using AI agents (programs that work on their own across several steps) | 12% of AI users | Muck Rack, State of AI in PR 2026 |
| Considering visibility in AI search at least somewhat important | 73%, while 29% say nobody in their organisation is responsible for it | Muck Rack, State of PR 2026, 1,115 respondents, May–June 2026 |
| Journalists using at least one AI tool | 82% (77% a year earlier) | Muck Rack, State of Journalism 2026, approx. 900 journalists |
| Journalists who immediately delete pitches outside their beat | 88% | Muck Rack, State of Journalism 2026 |
| Communicators worried that AI will amplify negative narratives about their brand | 64%; 36% have already encountered disinformation about their brand | We. Communications and USC Annenberg, over 600 respondents in the US, November 2025 |
What this means. The honeymoon phase is over: the share of users has stopped growing, but company policies, training and paid tools are multiplying fast. The industry has moved from asking “should we use AI?” to asking “how do we use it safely and with a measurable effect?”. The second lesson: journalists work with AI too, and they reject off-target pitches faster than ever — mass mailings generated by a machine work against the brand.
12 uses of artificial intelligence in public relations
The table below shows what the machine actually does in each use case, and what must stay with a person. This split is the simplest test of whether an implementation is safe.
| Use | What AI does | What stays with a person |
|---|---|---|
| 1. Media and online monitoring | Collects and groups mentions, removes duplicates, summarises | Judging what matters for the business |
| 2. Sentiment analysis | Classifies the tone of statements, shows changes over time | Checking irony, context and misclassifications |
| 3. Crisis early warning | Detects an unusual rise in mentions and estimates how fast it spreads | The decision to respond, the wording of the statement, legal sign-off |
| 4. Choosing journalists and media | Matches a story to what a given journalist actually writes about | The relationship, judgement, respect for the newsroom’s time |
| 5. Drafting press materials | Prepares drafts, headline variants and summaries | Facts, quotes, brand voice, accountability for the content |
| 6. Spokesperson preparation | Simulates tough questions and checks the consistency of answers | Attitude, credibility, the limits of what can be said |
| 7. Competitor analysis | Compares media presence, topics and share of voice | Strategic conclusions |
| 8. Brand visibility in AI answers | Checks what language models say about the brand and which sources they use | The publication and relationship plan that changes this picture |
| 9. Thought leadership content | Organises the team’s data and knowledge into arguments | An original thesis, experience, the expert’s signature |
| 10. Internal communication | Versions of messages for different employee groups, frequently asked questions | Tone in difficult situations, personnel decisions |
| 11. Reporting results | Pulls data from many sources into one picture | Telling coincidence apart from real impact |
| 12. Compliance checks | Flags missing labels, risky wording, personal data | The final decision to publish |
We have written more about the benefits and risks in our articles on the benefits of using AI in PR and marketing and on the challenges of AI in PR.
Brand visibility in AI: a new job for PR
More and more buying decisions start with a question asked to ChatGPT, Gemini, Claude or Perplexity rather than to a search engine. So the question is: where do these models get their knowledge about companies? The answer from the largest study to date is clear, and it is good news for PR.
According to Muck Rack’s study “What Is AI Reading?” from May 2026, covering more than 25 million links from ChatGPT, Claude and Gemini answers, 84% of cited sources are earned media (coverage the brand did not pay for: articles, research, public sources, independent websites). Paid content accounts for 0.3%. Journalism alone makes up 27% of citations, and that share has been stable across three consecutive editions of the study since July 2025. For questions about industry trends, the share of answers citing journalistic media reaches 46%.
In practice this means three things:
- You cannot buy your way into an AI answer with advertising. What earns that place is credible, independent media presence.
- Freshness matters. According to the same study, more than half of the cited journalistic articles are less than a year old, and after six months the number of citations drops noticeably.
- Each model behaves differently. ChatGPT attaches sources to 96% of its answers, Gemini to 82%, Claude to 55% — a visibility plan has to be checked separately for each of them.
That is why optimising for AI search (in industry terms: GEO, generative engine optimisation, and AEO, answer engine optimisation) is not only a job for the SEO team. To a large extent it is PR work: earning coverage in sources that the models consider credible, and making sure the facts about the brand are consistent everywhere.
How to increase your brand’s visibility in AI — five steps
- Measure the starting point. Ask the models 20–30 questions your customers ask, and record whether and how the brand appears.
- Check the sources. Which websites do the answers in your category rely on? Those are the targets of your media work.
- Get your facts in order. The name, scope of services, contact details and figures must be identical on the website, in profiles and in publications.
- Publish data, not claims. Models are more willing to cite something concrete: a study, a figure with a source, a definition.
- Repeat the measurement every month. Changes in AI answers take weeks, not days.
This is how we do this work for clients: AI search positioning — brand visibility in ChatGPT, Gemini and Perplexity, with a baseline measurement across four models and a plan for publications in the sources these models cite.
Rules and regulations: PRSA, FTC, the AI Act
US: PRSA ethical guidance
In October 2025 the Public Relations Society of America updated its “Promise & Pitfalls” guidance on the ethical use of AI. The key requirements: disclosure when AI has materially influenced content, a decision or contact with an audience; full accountability of the practitioner for every piece of work prepared with the help of AI; caution when entering confidential client data into public tools; vetting tool vendors for data protection; and ongoing human oversight of the process.
US: the FTC ban on fake reviews
Since 21 October 2024, a rule from the US Federal Trade Commission (FTC) has prohibited creating, buying and disseminating fake reviews and testimonials — explicitly including those generated by AI. Knowing violations can lead to financial penalties. For PR this is a clear line: AI can help collect and organise genuine customer reviews, but it must not invent them.
European Union: Article 50 of the AI Act
Since 2 August 2026, the transparency rules in Article 50 of the EU Artificial Intelligence Act (AI Act) apply across the EU, including Poland. The 2026 simplification package postponed deadlines for high-risk systems, but it did not postpone these obligations. Three matter most for communications teams: a chatbot must tell people it is an AI; images, audio and video that could pass for real (so-called deepfakes) must be labelled; and AI-generated text published on matters of public interest must be labelled or undergo documented human editorial review. Penalties reach EUR 15 million or 3% of worldwide turnover. A short transition period until 2 December 2026 applies only to technical content marking by providers of tools that were on the market before August.
This section is for information only and does not constitute legal advice. Consult a lawyer before putting these rules into practice.
How to measure the effects of AI in PR
AI makes counting easier, but it does not excuse you from measuring honestly. Three rules we stick to:
- We separate four levels according to the international AMEC standard (the International Association for the Measurement and Evaluation of Communication): outputs (what was published), outtakes (who saw and understood it), outcomes (what changed in attitudes and behaviour) and impact (what it delivered for the organisation).
- We do not use advertising value equivalency (converting coverage into the price of an ad) as proof of value — industry standards have rejected this measure for years.
- Coincidence is not impact. When traffic or sales rise after a campaign, we check it against a baseline from before the activity, a comparison group or a test in selected regions.
Implementing AI in a PR team in 90 days
- Days 1–15: audit. A list of the team’s tasks and the time they take. Choose 3–5 repetitive tasks where AI will save the most.
- Days 10–20: rules. A short policy: which tools, what data may be entered into them, who approves publications, how we label content.
- Days 15–45: pilot. One use case, e.g. monitoring and early warning, with time and quality measured before and after.
- Days 30–60: visibility in AI. A baseline measurement in language models and a plan for publications in the sources these models cite.
- Days 45–75: training. Hands-on workshops on the team’s own cases, not a generic course.
- Days 60–90: evaluation. The decision: scale up, improve or withdraw — based on numbers, not impressions.
What AI in PR will not do
AI will not build a relationship with a newsroom, will not take responsibility for a statement in a crisis and will not judge whether a given topic is appropriate on a given day. Nor does it know facts that are not in its data — although it can invent them convincingly. That is why a simple rule applies in our work: AI analyses and proposes, a person decides, and every figure in our material has a source. This matters most in a crisis — see how we work in crisis management.
How we work with AI at Commplace
Since 1996 we have been running public relations for companies across many industries. The PR AI module of the Commplace® AI ecosystem combines monitoring, analysis, early warning and measurement of brand visibility in language model answers — while decisions, media relations and accountability for every word stay with people. You will find examples of our work in our case studies, and more about our approach to communication in the article what PR is and how it affects business and on the page about communication strategy.
Let’s talk about what the media and AI models are saying about your brand today →
Frequently asked questions about AI in PR
What is AI in PR?
It is the use of artificial intelligence for media monitoring, sentiment analysis, preparing materials, choosing journalists and detecting crises early. Decisions, relationships and accountability for the content stay with people.
Will AI replace PR professionals?
No. It takes over repetitive and computational work, but not judging the situation, media relations or accountability for every word. According to Muck Rack (2026), 76% of PR professionals use generative AI — as support, not as a replacement.
How can I check my brand’s visibility in AI?
Ask ChatGPT, Gemini, Claude and Perplexity the questions your customers ask, record the answers and sources, and repeat the measurement every month. Specialised tools that monitor model answers are used for systematic measurement.
How can I increase my brand’s visibility in ChatGPT and other AI models?
Through coverage in credible, independent sources, consistent facts about the brand across the internet and data-based content. Muck Rack’s study from May 2026 shows that 84% of the sources cited by the models are earned media, and only 0.3% are paid content.
Does PR content created with AI have to be labelled?
In the EU, Article 50 of the AI Act applies from 2 August 2026: deepfakes, among other things, must be labelled, as must AI-generated text published on matters of public interest, unless it has undergone documented human editorial review. In the US, PRSA guidance recommends disclosure when AI has materially influenced the content.
What are the biggest risks of AI in PR?
Invented facts presented in convincing language, leaks of confidential data into public tools, mass off-target mailings to journalists, and breaches of the rules on fake reviews and content labelling.
How is GEO different from SEO?
SEO (search engine optimisation) competes for a place in the list of Google results. GEO (generative engine optimisation) works to make sure the brand appears, and is described correctly, in an AI model’s answer. In GEO the key is credible external sources — which means PR work.
Where should a PR team start with implementing AI?
With an audit of the team’s tasks and a short AI usage policy, followed by one pilot with measurement before and after. Organisations with such policies already account for 51% of PR professionals’ workplaces in the Muck Rack study (2026).
Sources
- Muck Rack, The State of AI in PR 2026 (survey, 564 respondents, December 2025).
- Muck Rack, The State of PR 2026 (survey, 1,115 respondents, May–June 2026).
- Muck Rack, State of Journalism 2026 (March 2026).
- Muck Rack Generative Pulse, What Is AI Reading? (May 2026).
- PRSA, Promise & Pitfalls: The Ethical Use of AI for Public Relations Practitioners (updated October 2025).
- We. Communications and USC Annenberg, report on AI and brand reputation (November 2025).
- FTC, rule banning fake reviews (in force since 21 October 2024).
- European Commission, transparency obligations under Article 50 of the AI Act (July 2026).
Author: Sebastian Kopiej, CEO of Commplace®. In public relations since 1996. The Polish original was prepared with the help of AI tools and editorially reviewed by the author; this English version was translated from it with the help of AI tools. Data current as of 22 September 2026.