Key takeaways
- AI assistants overwhelmingly cite non-paid, earned sources, and journalism is a major share of what they read.
- Press release citations in AI answers grew sharply in late 2025, and the releases that got cited share a recognizable structure.
- Links in distributed press releases aren't a ranking shortcut: Google treats keyword-stuffed release links as link spam. The value is the mention itself.
- Write releases like news: real facts, numbers, bullet points and a neutral tone.
For decades, a press release had one audience: reporters. You sent it out and hoped a journalist picked it up. Today a press release has a second audience that reads everything, never sleeps and answers millions of questions a day: AI assistants.
When someone asks ChatGPT or Gemini to recommend a business, the assistant looks for credible sources that describe the options. News sites are high on that list. So the question for business owners has shifted from "will a reporter notice this?" to "will this give AI something trustworthy to cite?"
What the research says AI is reading
Muck Rack has been studying which links AI assistants cite when they answer realistic questions. Its findings are consistent:
- In its July 2025 study, more than 95% of the links cited by AI came from non-paid sources, and roughly a quarter of the cited links were journalistic content.
- In its December 2025 update, based on more than one million cited links, about 94% of citations came from non-paid sources and earned media alone accounted for 82%. Journalism made up about 20 to 30% of citations over any period.
- The same update found that press release citations had increased five-fold since July 2025.
The takeaway is simple: AI assistants lean on what other people publish about a business, not on what the business says about itself. Earned coverage, the kind you can't simply buy as an ad, is the raw material of AI recommendations.
The press releases AI actually cites
Not every release gets picked up. Muck Rack's December 2025 report looked at the ones AI cited most and found a clear pattern. On average, the cited releases had:
- 2x as many statistics
- 2.5x as many bullet points
- 30% more action verbs
- A 30% higher rate of objective sentences
- More unique companies and products mentioned
That lines up with how AI assistants work. They're looking for facts they can lift and attribute: numbers, specifics and clear statements. A release full of adjectives ("industry-leading," "world-class," "passionate") gives them nothing to use. A release that says "opened a second location in Gilbert, adding 12 jobs and same-day service for 40,000 more households" does.
An honest word about links
Many people still buy press releases for the backlinks. That's the wrong reason, and it can backfire. Google's spam policies specifically list "links with optimized anchor text in articles, guest posts, or press releases distributed on other sites" as an example of link spam.
So the value of a distributed press release isn't a stuffed keyword link. It's the mention: your business name, what you do and where, published on sites that search engines and AI assistants already trust. That's exactly the evidence the research above shows AI relies on. We keep links in our releases natural, usually just your business name pointing to your website.
How to write a press release for the AI era
- Start with something true and new. A new service, a second location, a milestone, a hire, a community project or genuinely useful seasonal advice.
- Lead with facts. Who, what, where, when, and a number wherever you honestly have one.
- Use bullet points for the key details. They're easy for people to skim and easy for AI to quote.
- Keep the tone neutral. Write it like a reporter would, not like an ad.
- Use your exact business name and location the same way they appear on your website and Google Business Profile, so every mention connects back to you.
- Include a short quote from the owner that adds a real point of view, not a slogan.
Where it gets published matters
A release that sits only on your own website is just more of you talking about you. Distribution is what turns it into third-party coverage. Recognizable national names carry the most credibility with readers, and wide news networks spread the same story across dozens or hundreds of sites, creating the "mentioned everywhere" pattern AI looks for.
One nuance worth knowing: some outlets don't allow their press pages to be indexed by Google. Those placements still add credibility and give you an "as seen in" badge, but they won't show up in search results, so they shouldn't be your only channels. You can see how we label each channel in our distribution table.
For Google's AI Overviews in particular, the rule is the same as regular search: a page has to be indexed and eligible to show a snippet before it can appear as a supporting link. Indexed placements on trusted news sites are the ones doing the heavy lifting.
Make it a habit, not a one-off
One release is a moment. A steady rhythm of real stories builds a body of coverage that keeps growing, and AI answers change as new sources appear. That's why our plans include a press campaign every month, alongside the SEO foundation and AI visibility tracking that show whether it's working.
Curious which outlets would carry your story? Get a free AI visibility check and we'll walk you through it.
Our team plans and writes press campaigns, builds cloud stacks and runs AI visibility tracking for businesses every month. We check every claim in our articles against the sources listed below.
Sources
- Muck Rack (via MarTech Series): Muck Rack Study: Generative AI Relies Heavily on Earned Media and Journalism, July 2025
- Muck Rack (via MarTech Series): Earned Media Still Drives Generative AI Citations as Press Release Visibility Grows, December 2025
- Google Search Central: Spam policies for Google web search — link spam, including optimized anchor text in distributed press releases
- Google Search Central: AI features and your website
