The Press Herald scrutinized how The Maine Mirror reports. I responded by opening my evidence to its reporter and asking him to keep looking.

By Ryan Michaels | The Maine Mirror

On August 10, Portland Press Herald reporter Ethan Horton sent me a message on Facebook.

He said he was working on a story about Maine news sites “driven primarily by AI,” including The Maine Mirror, and wanted to speak with me about how The Maine Mirror sources and reports information and how artificial intelligence is used to present it.

(Screenshot of Portland Press Herald reporter Ethan Horton’s August 10 Facebook message requesting an interview with The Maine Mirror about its sourcing, reporting practices and use of artificial intelligence.)

I saw the message.

I didn’t respond.

There is an important reason why.

When I clicked the Facebook profile attached to the message, I saw a profile bearing Horton’s name with approximately 90 friends, no mutual friends visible to me, no profile photograph visible in the view I encountered and no immediately apparent information connecting the account to the Portland Press Herald.

(Screenshot of the Facebook profile Ryan Michaels encountered after receiving the August 10 message. The view displayed approximately 90 friends, no mutual friends visible to Michaels and no immediately apparent information identifying the account as belonging to a Portland Press Herald reporter.)

In a world of spam calls, automated messages, impersonation attempts and fake social-media profiles, I did not feel comfortable assuming that an unsolicited Facebook account claiming to represent a major newspaper was authentic simply because the message said so.

I could have independently verified Horton.

I could have searched for him, contacted the Press Herald or emailed the address he provided.

I didn’t.

Five days later, I learned the message had been genuine.

August 15: The Article Is Published

On August 15, the Portland Press Herald published Horton’s article, “Is the Maine news on your social media feed AI-generated? How you can tell.”

The article examines a collection of Maine websites and social-media operations through the broader subject of artificial intelligence in news.

And an important distinction needs to be made about what Horton actually reported.

His article contains some very serious findings involving other operations.

It reports that Twin Cities Media published a story containing “multiple errors, fabricated sources,” lists purported reporters whom the Press Herald says “are not real,” and elsewhere describes sources that appeared fabricated. The article separately describes a Portland Post story containing what appeared to be an entirely fabricated quotation and reporting an award nomination before the nominations had even been announced.

Horton did not accuse The Maine Mirror of those things.

He did not report that The Maine Mirror fabricated a source.

He did not report that The Maine Mirror invented a reporter.

He did not identify a fabricated quotation published by The Maine Mirror.

He did not demonstrate that one of The Maine Mirror’s investigations was hallucinated.

Those distinctions matter.

Yet The Maine Mirror appears within the same broader article and ecosystem. After discussing Twin Cities Media, Horton writes that beyond that operation there is the Portland Post and “the AI-driven Maine Mirror,” which he calls a “purportedly investigative site.”

The criticism directed specifically toward The Maine Mirror is substantially different.

Horton points to AI-generated imagery and characteristics of our writing he associates with artificial intelligence: bullet points, three-item lists, grammar, vocabulary and short paragraphs.

He describes me as a “prominent social media conservative” and later a “conservative commentator.”

And one of the central criticisms directed specifically at The Maine Mirror is transparency. Horton writes that I do not “openly discuss or disclose” the use of AI in publishing the site.

That is where I believe important context exists.

There were disclosures surrounding my use of artificial intelligence. At the same time, I could understand an argument that those disclosures could have been clearer, more standardized and substantially easier for a reader to locate.

So I had a choice.

I could spend my energy fighting over whether Horton had been fair to me.

Or I could make the transparency unmistakable and then ask him to examine something considerably more important.

I chose the latter.

I Took the Criticism Seriously Without Accepting the Entire Premise

On August 17, The Maine Mirror expanded its public editorial transparency through Behind the Mirror, a hub organized around three principles:

Our Standards. Our Methods. Our Accountability.

It includes foundational standards addressing:

BTM-001 — Artificial Intelligence & Editorial Transparency

BTM-002 — Reporting, Sourcing & Verification

BTM-003 — Editorial Accountability, Corrections & Transparency

That was not an admission that The Maine Mirror had fabricated journalism.

Horton never accused us of that.

It wasn’t an admission that using artificial intelligence was inherently improper.

I don’t believe that.

And it wasn’t an admission that every criticism in the Press Herald article was correct.

It was something much simpler.

If readers could benefit from clearer disclosure about how The Maine Mirror operates, I could provide it.

There is no contradiction between disagreeing with criticism and finding something useful within it.

Accountability also cannot be something The Maine Mirror demands exclusively from government agencies, elected officials and everyone else.

If someone scrutinizes us, they are entitled to scrutinize us.

If that scrutiny reveals an opportunity to make our methods more explicit, I can do that without surrendering the right to challenge the framing surrounding that criticism.

But after doing that, I wanted to move the conversation somewhere else.

Away from appearances.

Away from assumptions.

Away from whether a paragraph “sounds like AI.”

And directly into the evidence.

August 17: I Asked Horton to Keep Looking

At 10:45 a.m. on August 17, I emailed Horton.

The subject was:

“Following Up — An Invitation to Examine the Evidence.”

I explained why I hadn’t responded to his original Facebook message.

I acknowledged that I disagreed with some of his characterizations while also recognizing that parts of the criticism gave me something useful to consider.

Then I explained what I had done in response.

(Page one of Ryan Michaels’ August 17 email to Ethan Horton explaining the missed Facebook outreach, discussing The Maine Mirror’s expanded editorial standards and inviting further examination.)

But that wasn’t the real reason I emailed him.

The real reason appears further down.

I wanted Horton to audit my journalism.

Don’t Trust Me. Test Me.

I gave Horton a specific example.

I directed him to an August 12 Maine Mirror investigation concerning a recorded encounter involving Maine DHHS and what police were subsequently told about that encounter.

I provided the related NotebookLM Deep Dive.

And, most importantly, I provided access to the primary-source material underlying the reporting.

That created a remarkably straightforward test.

Horton could examine:

1. The primary-source evidence.
2. The Maine Mirror article derived from that evidence.
3. The AI-assisted NotebookLM presentation derived from those materials.

(Page two of the August 17 email showing the Maine DHHS investigation and NotebookLM Deep Dive provided to Horton, along with Michaels’ invitation to compare the primary evidence, The Maine Mirror’s reporting and the AI-assisted presentation.)

I didn’t ask Horton to believe my interpretation.

Quite the opposite.

I wrote:

“You don’t have to trust my characterization of the evidence.”

“You don’t have to trust The Maine Mirror.”

“You don’t have to trust NotebookLM.”

“You don’t even have to trust me.”

Then I asked him to do something:

Look.

Listen to the recording.

Examine the source materials.

Compare them with my article.

Compare them with the AI-assisted podcast.

Determine whether the reporting accurately reflects the evidence.

If I omitted important context, identify it.

If AI introduced unsupported information, expose it.

If my conclusions outran the evidence, challenge them.

And I made the commitment explicit:

“If you find something I got wrong, tell me.”

That wasn’t rhetorical.

It still isn’t.

If my journalism doesn’t survive examination of the primary sources, that matters.

But there is another possible result.

What if it does?

What if the source materials raise legitimate questions?

What if the reporting accurately describes them?

What if the evidence points toward conduct that deserves independent investigation by a newsroom with vastly greater resources than mine?

Then I hope a journalist follows that too.

That’s Where This Story Changes

This is the part of this story I find difficult to ignore.

The Portland Press Herald found The Maine Mirror interesting enough to examine.

A reporter looked at how our stories are written.

He looked at our images.

He examined characteristics of our formatting and language.

He looked at my personal social-media activity.

He considered how artificial intelligence might be involved.

He considered whether our disclosure was adequate.

He wrote about my perceived political identity.

All of that was apparently worth examining.

I’m not saying it shouldn’t have been.

I’m saying: keep looking.

Because once I knew the reporter was real, I didn’t hide from the scrutiny.

I opened the door wider.

I gave him the underlying evidence.

I gave him permission to challenge me.

I effectively said:

You looked at how I report. Now look at what I found.

What Happened After Publication Matters Too

According to the email tracking information available to me, my message was sent at 10:45 a.m. on August 17 and registered as opened approximately one minute later, at 10:46 a.m.

(Email tracking screenshot showing “Following Up — An Invitation to Examine the Evidence” sent August 17 at 10:45 a.m. and registered as opened by the recipient at approximately 10:46 a.m.)

That proves very little beyond establishing the sequence of events.

An “opened” notification does not establish that Horton carefully read the entire email.

It doesn’t establish that he opened every link.

It doesn’t establish that he reviewed the evidence.

It doesn’t tell me whether he forwarded anything to an editor.

And it certainly doesn’t prove that Horton or the Press Herald deliberately ignored anything.

I won’t make claims the evidence doesn’t establish.

As of this writing, however, I have received no response to that invitation that I can report.

Meanwhile, the Press Herald article continues to tell readers:

“The Maine Mirror did not respond to a request for an interview.”

That statement accurately described what happened before the article was published.

But someone encountering the article now would not learn what happened afterward.

They would not know that I subsequently contacted Horton.

They would not know why I had initially questioned whether the Facebook outreach was authentic.

They would not know that I directly engaged with his criticism.

They would not know that I made The Maine Mirror’s AI practices and editorial standards more explicit and accessible.

And they would not know that I invited Horton to independently audit the evidence underlying my journalism.

I am not claiming that every news article must be continuously rewritten whenever something happens after publication.

Nor am I claiming that the original statement was false when published.

I’m asking something narrower:

Would someone reading that article today have a materially fuller understanding of the story if they knew what happened next?

Readers can decide that for themselves.

The Bigger Question

The final portion of my email to Horton was titled:

“The Bigger Question.”

(Page three of the August 17 email, including “The Bigger Question,” where Michaels asks what it takes for evidence brought forward by ordinary Mainers to receive serious examination from larger Maine newsrooms.)

Because ultimately, this isn’t about whether Ethan Horton likes The Maine Mirror.

It isn’t about whether I like his article.

It isn’t even primarily about artificial intelligence.

It’s about what journalism chooses to look at.

I didn’t come through journalism school or spend years inside a traditional newsroom.

I’m an everyday Mainer.

I’m also a disabled veteran.

I came to this work after experiences with government left me asking questions that I believed deserved answers and after repeatedly trying to get institutions and people with considerably more resources than I have to look at evidence I believed mattered.

Eventually, I reached a simple conclusion:

Be the change you wish to see.

The Maine Mirror grew from that.

Technology—including artificial intelligence—gave one person capabilities that would have been extraordinarily difficult to possess even a few years ago.

I can organize hundreds of pages of records.

I can compare timelines.

I can work through government documents.

I can turn complicated source material into formats ordinary people can understand.

I can publish.

I can ask questions.

And I can show people my sources and invite them to determine whether I got it right.

That doesn’t make me infallible.

It doesn’t make every allegation I investigate true.

And it certainly doesn’t mean anyone should believe something simply because I published it.

Quite the opposite.

Don’t believe me because I’m a veteran.

Don’t believe me because I tell you I’ve been wronged.

Don’t believe me because The Maine Mirror published something.

Look at the evidence.

The Asymmetry Is What I Can’t Stop Thinking About

This is the surreal part.

A major Maine newspaper had enough interest in what I’m doing to scrutinize my use of artificial intelligence.

My images were worth examining.

My writing style was worth examining.

My disclosure practices were worth examining.

My Facebook presence was worth examining.

My perceived politics were worth mentioning.

Fine.

Journalists should scrutinize people making claims to the public.

But when the person being scrutinized responds:

Here are the records.

Here is the recording.

Here is what I reported.

Here is what the AI produced.

Compare them.

Try to prove me wrong.

And if the evidence raises serious questions about Maine government, investigate those too.

What happens then?

That’s the question.

Not:

Why won’t Ethan Horton believe Ryan Michaels?

I don’t want him to believe me.

The question is:

If The Maine Mirror was worth examining this closely, isn’t its evidence worth examining too?

“We Know to Ask Questions”

Near the end of Horton’s own article, Maine Press Association Board President Dan MacLeod describes what distinguishes reporters:

“We know to ask questions.”

I agree.

Ask them.

Ask me questions.

Ask whether The Maine Mirror uses AI.

Ask whether our disclosures are sufficient.

Ask where our information comes from.

Ask whether my characterization of a government record is accurate.

But don’t stop there.

Ask the agency questions too.

Ask why a recording appears to show one thing while another account describes something different.

Ask the officials involved.

Ask for their documents.

Ask whether I’m missing something.

Find contradictory evidence.

Try to disprove the allegation.

Do the work necessary to find out what happened.

If that investigation establishes that Ryan Michaels fundamentally misunderstood the evidence, publish it.

I’ll have to answer for that.

But if the evidence survives scrutiny?

Publish that too.

This Isn’t About One Reporter

I want to be careful here.

I don’t know why Horton hasn’t responded.

I don’t know whether he has examined any of the materials.

I don’t know what his editors have discussed.

I don’t know what assignments he has.

I don’t know what investigations the Press Herald may or may not be pursuing.

And I am not going to manufacture motives where I don’t have evidence.

That would undermine everything I’m arguing for.

Instead, I’m asking a question about journalism itself.

Every newsroom makes choices.

There are limited reporters.

Limited money.

Limited hours.

Thousands of possible stories.

Editors inevitably decide what deserves attention and what doesn’t.

So:

How are those decisions made?

What makes one person’s story worth investigating?

What makes another person’s records worth opening?

How much does institutional status matter?

How much does politics matter?

How much does audience interest matter?

How much does web traffic matter?

How much does legal risk matter?

How much does a source’s perceived credibility matter?

And how much does the existence of independently verifiable primary evidence matter?

I don’t know all of those answers.

But I think Mainers deserve to understand them.

Then The Maine Wire Raised Another Question

The conversation became more complicated when The Maine Wire published its own response to the Press Herald article.

The Maine Wire pointed to comments from Maine Trust for Local News board chairman Mark Stodder discussing ways AI could potentially help newsrooms cover the enormous number of public meetings occurring throughout Maine.

There is an easy response to that:

“See? They use AI too.”

I’m not interested in making that argument.

Because I don’t believe using artificial intelligence automatically discredits journalism.

If I did, The Maine Mirror wouldn’t use it.

The meaningful dividing line isn’t AI versus human.

It’s accountable versus unaccountable.

Was the information verified?

Are the sources real?

Does the evidence support the claim?

Is the role of AI represented transparently?

Is a human being ultimately responsible for publication?

Can readers distinguish reporting from illustration?

Will the publication correct something when the evidence establishes that it is wrong?

Those standards should apply to me.

They should apply to The Maine Mirror.

They should apply to the Portland Press Herald.

They should apply to The Maine Wire.

They should apply to anybody asking the public to regard their work as journalism.

Don’t Believe Me. Look.

The Portland Press Herald article put The Maine Mirror into a broader discussion that included some extraordinarily serious examples of what can go wrong when AI and journalism intersect.

Again, to be precise:

The Press Herald did not accuse The Maine Mirror of fabricating sources, inventing reporters or publishing fabricated quotations.

Its criticism of us centered principally on AI use, presentation and disclosure.

I responded to the part I believed could be made stronger.

I made our standards easier to find through the Behind the Mirror Hub.

Then I asked Horton to move beyond the surface.

I handed him something testable.

And that is ultimately what I want from journalism.

Not sympathy.

Not automatic credibility.

Not favorable coverage.

Scrutiny.

If you’re going to scrutinize me, scrutinize me completely.

Don’t believe Ryan Michaels.

Don’t disbelieve Ryan Michaels.

Look.

Open the documents.

Listen to the recordings.

Compare the evidence with what I wrote.

Contact the government.

Seek the missing context.

Try to prove me wrong.

And then tell the public what the evidence actually establishes.

Because if journalism is willing to spend time examining how an ordinary person is reporting, I hope it remains equally willing to examine what that ordinary person is trying to show it.

That’s why I created The Maine Mirror in the first place.

I am one person who encountered things I believed mattered and decided that waiting indefinitely for somebody else to look wasn’t enough.

So I started looking myself.

Now one of Maine’s largest news organizations has looked at me.

My response is simple:

Keep looking.

The door is open.

The sources are available.

The invitation remains.

And the question is bigger than either Ethan Horton or me:

When ordinary Mainers bring potentially consequential evidence to the institutions responsible for informing the public, what determines whether somebody looks?

---

Transparency Note: Consistent with The Maine Mirror’s published AI-transparency standards, artificial intelligence assisted with the organization, structure, refinement and editing of this article. The underlying events, arguments, source materials, editorial judgments and responsibility for publication remain Ryan Michaels’ own.

Support Independent Maine Journalism

If you believe journalism should follow the evidence, ask difficult questions, and give ordinary Mainers a chance to be heard, help The Maine Mirror keep doing exactly that.

Subscribe below for free to receive our latest investigations, reporting, and updates—and help strengthen independent, evidence-driven journalism in Maine.