For PIOs and communicators, the challenge is no longer just misinformation. AI is also changing how we verify authenticity, source and trust.

Artificial intelligence is moving fast. Public confidence in it is not.

That was my biggest takeaway from a new Quinnipiac University poll on AI. The numbers themselves are interesting, but for public information officers and communicators, the bigger question is what they mean for the way we do our jobs.

The poll found that 82% of Americans are concerned about artificial intelligence, 53% believe AI will do more harm than good in their day-to-day lives, and nearly three-quarters have little or no trust in the leaders of AI companies.

One result stood out to me even more. Americans are more concerned about people using AI to cause harm than they are about autonomous AI systems causing harm on their own. Fifty-two percent pointed to humans using AI as the bigger concern, compared with 33% who were more worried about autonomous AI agents.

That feels important for communicators, because the real issue is not just the technology. It is how people use it, how organizations explain it, and whether the public trusts the people behind it.

We Should Not Assume People Are Comfortable With AI

There are plenty of good reasons for communicators to use AI. I use it. I teach about it. I think people in this profession need to understand what these tools can do and where they can actually save time.

AI can help summarize information, organize notes, develop ideas, review documents, build scenarios and take some of the routine work off our plates. That can be useful.

But we should not confuse usefulness with trust.

According to the Quinnipiac poll, only 14% of Americans think powerful AI systems should continue developing at the current pace, and just 5% think development should move faster. Nearly half want development slowed while safety is evaluated, and another 30% want it stopped until those evaluations happen.

That tells me we need to be careful about how organizations talk about AI.

Calling something “AI-powered” may sound innovative internally. To the public, it may raise questions about privacy, accuracy, oversight or who is actually making decisions.

That does not mean organizations should stop using AI. It does mean we should be ready to explain how we are using it and why.

We Need to Be Able to Explain How We’re Using It

A lot of organizations are still working this out.

Some have detailed AI policies. Some have banned certain tools. Others are in that awkward middle ground where employees are already using AI, but the organization has not really decided what the rules are yet.

PIOs and communicators should be part of that discussion.

What information can employees put into these systems? What should never be entered? How are AI-generated answers checked? When should AI use be disclosed? Can staff create AI-generated images, audio or video? Is AI helping with decisions, or making them? Who is responsible when something goes wrong?

Eventually, somebody is going to have to answer those questions publicly.

The poll found that 91% of Americans believe AI guardrails are important and 86% support independent safety standards even if those standards slow development.

That is worth paying attention to. People are not just interested in whether the technology works. They want to know who is watching it and who is responsible for the outcome.

The More Convincing AI Gets, the More Careful We Have to Be

One of the things I worry about most with generative AI is not a bad answer that is obviously wrong.

It is a good answer that is mostly right.

A polished response can look authoritative even when one fact is wrong, one source is outdated or one detail has been invented. That becomes a very different problem when you are a PIO working under time pressure and that information is about to go out under your agency’s name.

If AI helps draft a news release during an emergency and one number is wrong, nobody is going to blame the software. If an AI-generated talking point includes something that was never verified, the public is not going to care which system produced it.

The organization owns it.

That is why AI literacy for communicators has to be about much more than learning how to write prompts. We need to know how to check the work, verify sources, protect sensitive information and recognize when a task should not be handed to AI in the first place.

The better these tools get, the more important our own judgment becomes.

AI Is Also Changing the Crisis Communications Environment

This is where I think the issue gets even more serious.

Imagine a major fire, officer-involved shooting, natural disaster, school incident or controversial government decision. Now add a fake audio recording of the chief, a fabricated screenshot, synthetic photos or a convincing AI-generated video.

And imagine that material starts circulating before the organization has even had time to understand what happened.

PIOs have dealt with misinformation and disinformation for years. AI makes that problem faster, cheaper and much more convincing.

At some point, the question will not just be whether a piece of information is true. It will be whether people can tell what information actually came from us.

That means authentication needs to become part of crisis communications planning.

Organizations should already be telling people where official information will be posted. They should have processes for monitoring false information and responding quickly when fabricated content appears. And communicators should be asking what they would do if a believable video, audio clip or document falsely appeared to come from their agency.

That is not something I would want to figure out for the first time during a crisis.

This Is Not an Argument Against AI

There is another part of the Quinnipiac poll that I think is important.

Sixty-nine percent of Americans said it is important for the United States to keep pace with China in developing AI systems. At the same time, 91% said guardrails are important. Quinnipiac found that 65% of respondents hold both views.

That is probably closer to where many people really are on AI.

They see the value. They understand that it is not going away. But they also want some reasonable boundaries around it.

I think communicators should approach AI the same way.

Use it. Learn it. Understand what it can do well and where it can get you into trouble. Put policies around it. Verify what comes out of it. Protect sensitive information. And keep a human being accountable for the final product.

The Quinnipiac poll surveyed 1,202 U.S. adults from September 24 through September 27, 2026, with a margin of error of plus or minus 3.5 percentage points.

For PIOs and professional communicators, I do not think the takeaway is that we need to convince people to trust AI.

I think the bigger responsibility is making sure they can continue to trust us.