Generative AI has changed the way we search for information online. When Google launched AI Overviews in 2024, it fundamentally altered how information reaches audiences. These machine-generated summaries, displayed above traditional search results, now determine what millions see first, and often, what they see exclusively.

Parallel to this, the AI-centric search engine Perplexity has grown in popularity, handling 780 million search queries last May alone. Even ChatGPT is being treated as a search tool, with nearly half of all prompts either seeking information on a person, company, thing, or practical how-to guidance.

The outputs of these tools aren't neutral, but rather generated by what already exists online. For businesses, this means that the coverage you receive matters.

According to MuckRack's recent "What Is AI Reading?" report, nearly all citations in generative AI responses come from unpaid media, particularly from owned or earned coverage. For PR professionals and their clients, this means their work has a longer lifespan.

Earned coverage now carries influence beyond the reader. It feeds generative AI tools used by billions of people each day, shaping how brands are explained, summarized, and surfaced over time. What’s secured today can continue to inform AI-generated responses for months—potentially years.

PR is the new infrastructure

The focus has moved beyond individual placements. PR now centers on building a persistent information architecture that AI systems can reference, verify, and trust. While a strong article in a top-tier publication certainly still matters, PR teams need to take a more holistic approach.

Every public-facing piece of content you produce or secure is visible to these AI tools, and the challenge is to make sure your client's perspective appears consistently enough to register as credible.

Timing matters

One of the most striking findings in the report is how recent most AI-cited content is. Half of all sources referenced by generative AI were published within the past 11 months, with a small but meaningful share coming from the previous seven days. Older coverage can still appear, but its influence drops quickly.

This creates a visibility decay that's faster than most annual planning cycles can accommodate. Brands that communicated their positioning six months ago and have been quiet ever since are already fading from AI-generated responses.

Generative AI tools have a preference for new, relevant, and fresh content. This means visibility isn't something you can bank. Maintaining influence over how AI models understand a client's brand, requires a steady output of new, authentic content. Consistency isn't a virtue in this environment—it's a requirement.

Long live the press release

This preference for fresh content also explains why AI systems are increasingly sourcing information from press releases, with its share of citations multiplying by a factor of five between July and December 2025.

Press releases were long treated as background noise—something that was mainly useful during a campaign, with limited utility beyond that point. Now, they're treated as structured, verifiable source material that AI models trust.

As I recently told Inc., press releases act as a reference point when models try to confirm facts or fill in gaps. Without that reference point, the system is forced to look elsewhere, pulling outdated information or flattening nuance.

Fixing this is a lot harder than simply getting it right the first time.

Essentially, PR teams now serve two different audiences simultaneously. Human readers evaluate credibility, narrative, and emotional resonance. AI systems evaluate recency, consistency, cross-source verification, and domain authority.

The uncomfortable truth is that focusing on just one is no longer viable. As mentioned, a well-crafted feature in a prestigious outlet matters, but only if it's reinforced by structured data in press releases, validated by expert commentary, and updated frequently enough that AI models treat it as current.

Publication choice also matters differently. Reach alone is no longer sufficient. AI systems place greater value on subject-matter authority, which means visibility within the right industry environments matters as much as prominence.

PR post-AI

Generative AI has changed what PR teams do, but the core discipline remains unchanged. We're still shaping perception over time. What's changed is the audience, and campaigns need to consider both human audiences, as well as the AI algorithms that decide which brands deserve to get mentioned, and which get ignored.

The teams that will succeed in this environment are those that treat AI visibility as a strategic discipline, not an afterthought. This means consistent output across earned, owned, and structured content. It means optimizing not just for reach but for algorithmic trust. And it means recognizing that in an AI-mediated information landscape, inconsistency is indistinguishable from invisibility. Brands that communicate sporadically, no matter how compelling those individual moments are, will fade from the answers AI generates.

Generative AI isn't replacing the fundamentals of strategic communications. It's raising the stakes for getting them right.