AI can reduce time spent on research, drafting, organization, and monitoring, but it does not make a weak story newsworthy or replace journalist relationships. The practical use case is to let AI handle defined, reviewable tasks while people retain responsibility for facts, judgment, tone, and outreach.
Muck Rack’s 2025 State of AI in PR research found that 75% of PR professionals were using generative AI. Among users, 93% said it made work faster and 78% said it improved quality. The useful question is no longer whether AI can produce words; it is where AI improves a PR workflow without lowering relevance, accuracy, or trust.
Why PR Teams Are Using AI
The same Muck Rack research found that common PR uses included brainstorming at 82%, drafting at 72%, and research at 59%. Adoption has moved faster than governance, however: more than half of respondents said their organization still lacked a formal AI-use policy.
Speed is only useful when the result remains relevant and accurate. Cision’s 2025 State of the Media research found that 86% of journalists reject pitches that do not match their beat or audience, while 72% were concerned about factual errors in AI-generated PR content. AI should therefore improve preparation and review—not increase the volume of generic outreach.
Where AI Helps—and Where It Does Not
AI is most useful when the input, output, and review standard are clear.
– Research: Summarize approved background material, organize recent coverage, and suggest reporters to investigate. A person should still read the reporter’s recent work and confirm that the proposed story fits.
– Ideation: Generate possible headlines, story angles, objections, and questions. The communications team must decide whether the underlying news is genuinely timely and relevant.
– Drafting: Produce working drafts and variations from verified facts and approved messaging. Every statistic, quotation, name, date, and link still requires review.
– Monitoring: Group coverage, surface recurring themes, and prepare summaries. Automated sentiment or visibility scores should be treated as directional signals rather than proof of influence or causation.
AI does not replace relationship-building, editorial judgment, source verification, crisis judgment, or the ability to recognize when a story is not ready to pitch. It also cannot guarantee that a journalist will respond or publish coverage.
Your 4-Step AI-Powered Media Strategy
Step 1: Build Smarter Media Lists
AI can help organize a research queue by topic, outlet, geography, and recent coverage. Treat that queue as a starting point rather than a finished media list. Beat assignments change, databases become stale, and a plausible-looking contact can still be wrong.
Relevance matters more than list size. Cision’s 2025 research found that 86% of journalists reject pitches that do not align with their beat or audience. Use AI to narrow a research queue, then verify each contact against current bylines, outlet responsibilities, and stated pitching preferences.
Step 2: Draft More Relevant Pitches
Use AI to create a small number of pitch options from verified facts, approved messaging, and the journalist’s current coverage. Then select and revise the strongest version. Personalization should explain why this story fits this person’s audience; it should not imitate familiarity or invent a relationship.
Evidence check: Propel’s Q2 2024 Media Barometer analyzed 405,964 pitches sent during the first quarter of 2024 and recorded a 3.43% response rate. Pitches between 51 and 150 words performed best in that dataset. This does not prove that AI improves response rates; it supports keeping outreach brief, specific, and relevant even when AI assists with drafting.
Step 3: Manage Follow-ups Without Becoming Spammy
Automation is most useful for reminders, status tracking, and preparing follow-up drafts. A person should decide whether a follow-up is warranted, confirm that the reporter is still relevant, and approve the final message. If the same follow-up could be sent to any journalist covering any story, it is too generic.
Step 4: Monitor Coverage and Business Signals
AI-assisted monitoring can classify mentions, identify themes, and summarize coverage. Treat sentiment and AI-visibility scores as directional signals rather than proof of influence. Evaluate them alongside placement quality, message pull-through, referral traffic, branded search, qualified inquiries, and other outcomes tied to the campaign.
A Muck Rack analysis of more than one million links cited by Gemini, Perplexity, Claude, and ChatGPT from July through December 2025 found that 82% were earned-media sources. That figure applies to the models, links, and period studied; it does not mean 82% of every AI answer comes from earned media. It does support treating credible third-party coverage as one component of an AI-visibility strategy.
NisonCo’s AI Consulting team blends deep PR expertise with cutting-edge AI knowledge to help you secure more high-quality media coverage, learn more.
Essential AI Tools for Modern PR
Not sure how to convert a business development or sales idea into a credible media angle? Start with NisonCo’s AI PR Pitch Generator, then verify the facts and tailor the output before using it.
Where ChatGPT Work Fits in a PR Workflow
ChatGPT Work is designed for tasks with a clear outcome that may require multiple sources, tools, or steps and produce a reviewable deliverable. In PR, that makes it useful as a workflow and preparation layer—not as an automatic media-placement engine.
– Turn approved source files, research, and messaging into a reviewable campaign brief.
– Compare potential media opportunities in a structured review file before a person approves the target list.
– Create or refresh an editorial, announcement, or outreach tracker from approved information.
– Prepare a recurring monitoring summary from connected sources when those sources and permissions are available.
– Draft pitch options, briefing documents, talking points, and follow-up materials for human review.
OpenAI distinguishes these research, analysis, workflow, and deliverable tasks from Codex’s software-development role. Which information Work can access—and whether it can draft, write, share, schedule, or execute—depends on the connected systems, configuration, and permissions available to the user.
OpenAI’s Work administration guidance says source-system permissions remain in effect and recommends narrow permissions, approvals, and human review for high-impact actions. Contacting journalists, publishing material, and scheduling external communications should remain behind explicit human approval. ChatGPT Work does not guarantee journalist interest, responses, or earned coverage.
Human Review and Ethical Guardrails
Generative AI can produce language that sounds confident even when a fact, quotation, citation, or case example is wrong. The person or organization sending the material remains responsible for it.
– Verify every statistic against the original source and confirm that the date, population, and study scope match the claim being made.
– Never attribute an AI-generated quotation to a real person. Obtain approval for quotations and confirm names, titles, dates, and company descriptions.
– Give the model verified source material instead of asking it to invent supporting facts, examples, or industry data.
– Do not place confidential client information, embargoed news, personal data, or privileged material into a tool that has not been approved for that information.
– Review every media contact against current coverage. AI-generated media lists can contain outdated roles, irrelevant contacts, or fabricated information.
– Follow the disclosure requirements established by the outlet, client, employer, contract, and applicable law. There is no universal disclosure rule that can be attributed to every journalist or to the Associated Press.
– Keep a person responsible for the final story angle, factual review, personalization, sending decision, and response to the journalist.
The PRSA Code of Ethics emphasizes accuracy, honesty, transparency, and accountability. Those duties still apply when AI helps create or organize the work.
Start with One Reviewable Use Case
Choose one narrow task, such as organizing approved research, creating pitch variations from verified facts, or summarizing coverage. Define what a good output must contain, who reviews it, and which facts require source checks.
Measure whether the workflow saves time without increasing corrections, irrelevant outreach, or compliance risk. Expand only after the team has a repeatable review process and clear responsibility for the final work.
For a focused drafting starting point, try NisonCo’s AI PR Pitch Generator. For help designing a broader human-reviewed workflow, explore NisonCo’s AI consulting services and public relations services.
