There is a camera on the side of the road. You can see it. After more than a year of writing about artificial intelligence, algorithms, Big Tech, government databases, copyright and surveillance, I am beginning to think that cameras may be the least complicated part of this story. I initially approached many of these subjects as separate issues. The deeper I have gone, the harder that distinction has become to maintain.

They keep leading back to the same questions: Who has our information? Who controls it? Who can search it? Who decides what information reaches us? What did we actually consent to? What happens when the government wants information collected by private companies? And what happens when artificial intelligence stops simply retrieving information and begins drawing conclusions of its own? Ultimately, all of those questions lead to a much simpler one: Where do we draw the line?

Virginia has begun trying to answer that question. Del. Marcus Simon, responding to my recent News-Press reporting on automated license plate readers, framed the issue in a way that may ultimately define this five-part investigation.

“The News-Press’s reporting raises an important question that I don’t think we should dismiss: whether any amount of regulation can make a comprehensive database of people’s movements — and all the information it contains — truly safe from abuse,” Simon wrote.

That question goes well beyond whether Flock Safety works or whether a license plate reader can help police find a stolen vehicle. The larger question is whether there are technological capabilities that become dangerous simply because they exist — and whether regulation can ever completely eliminate that danger.

Virginia Draws Its Line

License plate readers are not new. What has changed is the scale at which they can operate, their declining cost and the ability to connect information across increasingly large networks.

“As the technology became cheaper, more powerful and increasingly networked through private companies like Flock,” Simon wrote, “it became possible to create something very different: a searchable record of where thousands or even millions of people have traveled, even though the overwhelming majority of them were never suspected of doing anything wrong.”

Virginia’s State Crime Commission recognized the problem after studying automated license plate readers in 2024. Its work led to HB 2724 during the 2025 General Assembly session, establishing statewide restrictions governing law enforcement’s use of the technology. Searches must be connected to legitimate law-enforcement purposes, audit records must be maintained, ordinary plate data generally must be destroyed after 21 days, and restrictions apply to sharing information. Intentional misuse can result in criminal penalties.

Virginia had drawn a line, but drawing a line and keeping everyone behind it are two very different things.

“Passing rules and making sure those rules are followed are two different things,” Simon wrote. When the Crime Commission followed up, Simon said some Virginia law-enforcement agencies were not complying with the requirements, while nearly a third did not respond to the Commission’s survey.

“That’s a pretty clear reason for continued oversight,” he said.

Georgia Shows Why That Matters

If anyone thinks misuse is merely hypothetical, recent events in Georgia deserve attention.

In July, the Georgia Bureau of Investigation arrested five former Albany Police Department officers after an internal audit allegedly found they had retained Flock license plate information multiple times for non-law-enforcement purposes. Later that month, a Conyers Police Department dispatcher supervisor was arrested after investigators alleged she accessed Flock more than 30 times for non-law-enforcement purposes. Other cases followed, including four former Savannah Police Department employees arrested in August following another internal audit.

The charges are allegations and do not establish guilt. But the number of investigations matters, and so does another fact: Auditing helped uncover several of them. That supports Virginia’s decision to require audit trails, but it raises another question that deserves attention: How many people should have access to this information in the first place?

The public backlash has become visible as well. A Flock camera in Lumpkin County, Ga., was vandalized and apparently thrown into the woods this summer, while protesters have demonstrated against surveillance technology. Destroying property is not a legitimate response to a policy disagreement, but dismissing the anger would miss something important. People are becoming increasingly aware of how much infrastructure capable of documenting their lives has been constructed around them, and some clearly do not trust the institutions controlling it.

This Didn’t Start With Flock

For me, the questions raised by these cameras did not begin with Flock. They have been building through much of my reporting over the past year.

In September 2025, I wrote “Analysis: Profit-Driven Algorithms Are Killing Our Society,” examining the invisible formulas increasingly deciding what appears on our screens and what happens when engagement — anger, fear, shock and outrage — becomes the currency determining which information rises and which disappears. I argued then that these systems do not simply reflect society; they increasingly help shape it.

In “The End of Freedom: The Titans Who Own America’s Reality,” I approached the issue from another direction, looking at the extraordinary power increasingly concentrated among a relatively small group of technology companies controlling much of the infrastructure through which Americans communicate, consume information and experience the world.

Then came artificial intelligence.

In March, I wrote “Beyer’s AI Transparency Bill Highlights Gaps in White House Plan on Creators and Copyright,” examining Rep. Don Beyer’s AI Foundation Model Transparency Act and the questions surrounding AI, copyright, creators and transparency. That story focused on what information goes into increasingly powerful AI systems and whether creators and the public can understand what is happening inside models that increasingly influence real-world decisions.

One line from that piece has stuck with me throughout this reporting: People cannot protect what they cannot see.

In May, in “The More Artificial the World Gets, the More People Crave Something Real,” I approached the technological transformation from the human side — what happens to communities, journalism and ordinary relationships when algorithms, automation and artificial intelligence increasingly mediate everyday life. I wrote then that the internet largely won because it was cheaper and more convenient, not necessarily because everything it replaced was inferior.

At the time, I saw these as different stories. Looking back at the reporting now, I’m not sure they were. They increasingly look like different pieces of the same story.

Flock is simply where some of those pieces become visible.

The Watchdogs Are Working

Other journalists are asking many of the same questions. The Intercept, WIRED, 404 Media, and other investigative and technology-focused publications have reported on surveillance systems, digital communications, location information, government access to commercially collected information, and the growing technological infrastructure available to law enforcement.

The Intercept’s reporting this summer on Signal messages and federal investigations surrounding anti-ICE protests in Minneapolis is one example. Researchers and civil-liberties organizations are digging as well, including the Brennan Center for Justice, which has examined how artificial intelligence and data-fusion technology could allow authorities to combine information from vehicle sightings, public records, commercial sources and other databases.

The work is being done. The harder question is whether people are seeing it.

That question goes directly back to the algorithms I was writing about last year. Data scientist Sahin Ahmed raised a version of it in a 2024 Medium essay, “Are You Really in Control of What You See Online, or Are Algorithms Controlling It for You?” Ahmed described algorithms as invisible gatekeepers curating what users encounter according to behavior, interactions and preferences.

The essay itself is not an academic study, but the question behind it is increasingly supported by research. A 2026 study published in Nature randomly assigned active U.S. users of X to algorithmic or chronological feeds for seven weeks. Researchers found that switching on the algorithm increased engagement, shifted some political opinions toward more conservative positions, and promoted conservative content while demoting posts from traditional media.

The politics of that particular result are almost beside the point. A different algorithm could produce a different outcome. What matters is that researchers demonstrated that an algorithm could alter the information people encountered and measurably affect some attitudes.

That matters enormously for journalism.

A reporter can investigate a story and a newspaper can publish it, but neither necessarily controls whether a reader ever sees it. Search engines rank results. Social-media platforms rank feeds. Recommendation engines determine what comes next. Increasingly, an invisible layer of technology sits between the person producing information and the person attempting to find it.

There is no evidence here to declare that technology companies are deliberately suppressing journalism critical of them, and I am not making that accusation. But there is every reason to examine what happens when investigative journalism about powerful technological systems increasingly depends upon other algorithmic systems to reach the public.

We built these systems because the amount of information available to us became impossible to navigate manually. Now those same systems increasingly help decide which pieces of information receive our attention.

The Camera Is Only One Piece

This is where Flock begins looking less like an isolated controversy and more like one visible piece of something much larger.

A license plate reader records a vehicle. A smartphone can generate another category of information. Social media creates another. Public records provide another. Commercial databases may contain still more. Historically, connecting thousands or millions of those records required enormous amounts of human labor. Modern computing has made that easier, and artificial intelligence is accelerating it again.

AI also introduces a fundamentally different problem because a machine no longer has to simply retrieve information. It can analyze information and attempt to draw conclusions from it.

A license plate reader may accurately record my vehicle on a road at 8:42 a.m. It does not know whether I was driving, why the vehicle was there, who was inside or what anyone was doing. But connect that record with enough additional information and a computer can begin identifying patterns and associations.

A collection of accurate facts does not necessarily produce an accurate conclusion.

That is particularly important because artificial intelligence can be wrong. The National Institute of Standards and Technology has warned about AI “confabulation” — confidently generated erroneous or false information — and the particular risks created when such systems are incorporated into consequential decision-making.

A chatbot getting an answer wrong is one thing. A machine reaching a wrong conclusion about a real person inside a law-enforcement, employment, financial or government system is something entirely different.

For decades, the privacy question has largely been: What information do you have about me?

We may increasingly need to ask another: What has your machine concluded about me from everything else it knows?

What Did We Agree To?

There is an important difference between government surveillance and private technology. The government can investigate, arrest and prosecute, and I do not get to click “Decline” when my vehicle passes a police camera.

My phone is different. I bought it. I downloaded the applications. I created the accounts. Somewhere along the way, I clicked “Agree.” That is a choice, but whether it constitutes an informed choice is another matter.

What does Google actually know about our movements? What does Apple know? What does Meta know? What information is retained and for how long? What is shared? What comes from data brokers? What can algorithms infer from it? And what happens when the government wants it?

I don’t know the answers to all of those questions. That’s precisely why I’m asking them.

Those are not accusations against Google, Apple, Meta, Flock or anyone else. Maybe consumers have considerably more control than many of us realize. Maybe companies have erected strong safeguards. Maybe government access is far more restricted than people fear. Or maybe the rabbit hole goes considerably deeper.

We’re going to find out.

Virginia Has Drawn One Line

Simon does not argue that police should abandon modern technology. “I don’t think the answer is to pretend these tools have no legitimate law-enforcement uses,” he wrote. “They can help find a stolen car, locate a missing child or provide an important lead in a serious criminal investigation.”

That distinction matters. Technology is not inherently the enemy. I use Google. I carry a smartphone. I use social media. I use artificial intelligence. Most of us participate in these systems because they have made our lives easier in extraordinary ways.

But convenient and harmless are not necessarily the same thing.

“There’s a huge difference between using technology to investigate a crime and building a system capable of tracking where everyone goes just in case that information might someday be useful,” Simon wrote.

He said Virginia’s work cannot end simply because new safeguards exist. “We need to see how these protections work in practice, whether companies and law-enforcement agencies are complying with them, and whether these massive, interconnected databases create risks that regulation alone can’t adequately address.”

Massive, interconnected databases.

That may ultimately be the phrase that defines this investigation.

The camera collects one piece. A database stores it. A phone generates another. A commercial service may possess another. Social media creates another. Algorithms influence which information reaches us, and artificial intelligence is increasingly capable of taking information flowing in the opposite direction and attempting to tell someone what all of it means.

Governments change. Administrations change. Companies change ownership. Employees come and go. Policies change. Databases are breached. Algorithms are rewritten. Technology created for one purpose finds another.

“Our laws need to keep up with technology,” Simon wrote. “But we also need to keep asking the harder question: are there some surveillance capabilities that go too far, even if they can be regulated?”

That is the question this series is going to pursue.

Virginia has drawn one line. Georgia is demonstrating why safeguards matter. Researchers are showing that algorithms can influence the information reaching us. Investigative journalists are uncovering increasingly sophisticated surveillance capabilities. Artificial intelligence is changing what can be extracted from all of that information.

None of those stories alone explains what is happening. Put them together, however, and something much larger begins to emerge.

Over the next four parts, the News-Press is going to follow the information: who collects it, who buys it, who can access it, what we actually agreed to, what government can obtain, what algorithms decide we see, what AI can infer and what rights ordinary people have when the technology gets it wrong.

I don’t know where all of those answers will lead. That’s the point of reporting. What is already clear is that the camera along the road is only one visible piece of a much larger system, and understanding that system is going to require looking beyond the camera and following the information wherever it goes.

The question is no longer whether this technology is coming. It’s here. The question is how much power we’re willing to give it — and whether we can ever take that power back.