If the last two editions of SAGA Signals were about AI’s bills coming due, this one is about what AI is doing to demand. PR agencies have a real moment in front of them as GEO puts earned media in vogue. Public sentiment toward AI is souring and politicians seem to be signaling a greater interest in “helping.”
The PR and communications opportunity
The clearest articulation of the PR moment came from Priceline’s head of communications. Christina Bennett told PRNEWS at the Meltwater Summit that GEO is giving PR teams a clear metrics-backed value proposition. She is benchmarking editorial citation share month over month and using LLMs to audit how the brand is perceived in earned media. She underscored that AI tools aren’t just task engines when she suggested you ask an LLM to assess your PR strategy. She makes the point that since the bulk of LLM citations are coming from editorial and owned media, it makes PR one of the most measurable inputs into AI discovery.
PR Daily made the same case from the supply side. Allison Carter’s piece argued that GEO has given media relations an unexpected second life. Muck Rack data shows that 27% of all AI answers cite journalism, jumping to 49% when the question requires a recent answer. Carter’s important caveat is that the media industry itself is in a rough patch, with challenges not just to their economics, but also press freedom.
The agency-economics version of this story is showing up in M&A conversations. O’Dwyer’s published a detailed argument from longtime agency M&A advisor Rick Gould that AI capability is becoming a central factor in PR firm valuations. Buyers are now asking whether a firm is simply selling hours or leveraging technology that multiplies output and insights. Firms that demonstrate consistent, AI-driven margin improvement can command a premium. Firms that do not invest will look antiquated to buyers within a few years.
Why it matters: PR agencies have a greater opportunity today than at any other point in recent years. Here’s why.
If a meaningful share of AI answers is coming from earned media, then the work your team has always done is the most strategically important input into how your clients show up in AI search.
At the same time, the M&A conversation is shifting. If you are within five years of a potential transaction, the AI question buyers will ask you will be much sharper.
That said, you can’t celebrate without taking action. Yes, there’s opportunity, but many PR agencies will need to rethink, retool, and reinvent in order to fully take advantage.
Don’t fumble it by falling into the performative AI trap
The most blunt counter to all of the above came from PR News. Their piece argued that AI in PR has become one of the most overused and misunderstood concepts in modern communications, and that most agencies are using it wrong. The article’s core argument is that performative AI use, AI for the sake of AI, and bragging about AI without corresponding results, are all undermining agency credibility with the clients and prospects they’re trying to impress.
Provoke Media’s perspective piece makes the craft argument. The thesis: original ideas and original research will continue to differentiate, even in the age of generative AI. The danger is letting AI replace what you do rather than allowing AI to help you do what you already do better. The first path leads to commodity output and eventual irrelevance. The second leads to real results.
A new white paper makes the practical version of the same point. EIN Presswire’s research report on what veteran journalists actually want from press releases in 2026 is a reminder that even before AI, technology had already enabled a rising tide of slop directed at journalists from agencies and communicators. AI now makes that slop cheaper to produce. The agencies that resist the urge to flood reporters with AI-generated noise and instead invest in understanding what specific journalists actually want will keep getting real results.
Why it matters: The agencies that get this moment right will use AI to do their existing work better — sharper targeting, better research, faster iteration on stronger ideas. The agencies that get it wrong will use AI to produce more volume in less time without any added value.
The job apocalypse isn’t inevitable, but the public mood is souring
Ezra Klein’s latest column provides a nice counter-narrative to the popular theme of AI doom. His argument focuses on the idea that something is always scarce, and when automation makes one thing cheap, demand shifts toward what cannot be automated. Klein notes that despite years of AI fearmongering, the unemployment rate in March 2026 was 4.3%, almost identical to the 4.4% rate in March 2020. This suggests we’re not even seeing the start of a trend toward AI-driven job loss at a macro level.
As AI mimics what humans can do on a computer, demand shifts toward the things AI cannot do, especially relationships, judgment, and the willingness to take real responsibility for an outcome. We have been here before and seen radical technology and job shifts – and survived. For example, the Neuron points out that farming represented 40% of US jobs in 1900 and is under 2% today.
A new round of polls makes the AI sentiment trend hard to dismiss. Axios summarized recent polling: 70% of Americans think AI is moving too quickly, with 68% of Republicans and 77% of Democrats agreeing. Only 18% of young people aged 14 to 29 say they feel hopeful about AI. Negative views of AI have risen from 34% three years ago to just over 50% today. A record number of data center projects were canceled in Q1 2026 amid local resistance. A graduation speaker was booed for calling AI “the next Industrial Revolution.”
Politicians have sensed the wind direction. The White House is reportedly considering an executive order that would create a vetting process for AI models before they are publicly released, though it has reportedly been put on pause (at least for now). The catalyst was the Mythos model, with administration officials worried about political fallout from a possible AI-enabled cyberattack. The proposal is a notable reversal from the same administration that revoked Biden’s AI safety executive order within hours of taking office in January 2025.
Why it matters: Agency owners are now operating in three overlapping audiences with very different views of AI. Your team has its own views. Your clients have theirs. And the general public is increasingly skeptical (if not becoming hostile). That divergence has practical implications.
The AI apocalypse and AI hate narrative threatens to undermine the progress that this new technology and tools permit. Clients, employees, and owners may make decisions based on emotional reactions.
Of course, AI is not all sunshine and unicorns. Agencies that articulate a more grounded view of what AI does and does not do will look smarter and earn more trust. This thinking becomes a useful lens for your own positioning. If AI commoditizes execution, what scarce thing does your agency produce? If the honest answer is “nothing in particular,” the AI conversation is the least of your strategic problems.
Two new realities to track
OpenAI opened ChatGPT ads to self-service. OpenAI launched a self-serve advertising platform for ChatGPT, available to U.S. advertisers of all sizes after previously requiring $50,000+ minimums. The platform is targeting $2.5 billion in ad revenue this year and $100 billion by 2030. Early data from AdClarity suggests average monthly ad spend is already around $109 million, three months into the pilot. Targeting works through natural-language context hints rather than traditional demographics. Conversion rates from LLM platforms are reportedly running at 1.5x other referral channels.
Why it matters: We don’t know where LLM-based ads are headed, but it’s worth trying a small test with projects where the fit is right. Learn the platform before clients ask you about it.
The NCSC warned of an AI-driven “vulnerability patch wave.” The UK’s National Cyber Security Centre issued a warning that AI-powered vulnerability discovery (driven by models like Mythos) will trigger a wave of urgent software updates across the technology stack. Mozilla already released 271 vulnerabilities in Firefox last week, all discovered using Claude Mythos, up from 22 in the previous Claude iteration. CISA is reportedly considering shortening federal patch deadlines from three weeks to three days.
Why it matters: AI security risk is now a category your clients are going to start asking about, especially in regulated industries.
You’re going to need to be more on top of the tools that you and your team are using and understand how to keep them up-to-date properly.
At the same time, client needs will likely pop up. Cybersecurity is becoming a more frequent communications topic, not less, and the AI vulnerability story is going to drive a lot of crisis and proactive comms work. Agencies positioned to help clients navigate this credibly will find new work. Agencies that treat security as someone else’s problem will be late to a conversation their clients are already having.
The bottom line
We covered a lot of ground in this edition of SAGA Signals, so I don’t blame you if your eyes glazed over a bit and you skipped to the end. If you did, here’s the TLDR and key takeaways:
- GEO puts earned and owned media expertise in the spotlight. Lean into the opportunity for LLM visibility with clients today, but don’t oversell what’s still an emerging discipline. The agencies that win will be the ones integrating AI planning and measurement before their competitors do.
- Performative AI use will undermine your credibility faster than no AI use. Your clients can tell the difference between agencies using AI to make their work better and agencies bragging about AI to mask the same old work (and perhaps attempt to artificially prop up fees).
- Public sentiment toward AI is shifting faster than many realize. Calibrate your client messaging and your own positioning to a market that’s more skeptical than it was six months ago. And ask yourself the harder question: if AI commoditizes execution, what valuable thing does your agency actually produce?
