AI’s expanding frontier and the data it demands
Artificial intelligence is rapidly changing how we live, work and govern. In public health and public services, AI tools promise more efficiency and faster decision-making. However, there is an increasing imbalance beneath the surface of this transformation. The ability to collect data outweighs the ability to control responsibly.
This goes beyond the technical challenges that will lead to a privacy crisis. From predictive policing software to monitoring tools and automated license plate readers, data about individuals is accumulated, analyzed and acted at an unprecedented rate. Still, most citizens don’t know who owns their data, how it is being used, or whether it is protected.
I saw this up close. As a former FBI Cyber Special Agent and now a leading CEO of a public safety technology company, I have worked in both the government and the private sector. One thing is that if you don’t modify the way data privacy is handled, AI will only exacerbate existing issues. And what is one of the biggest problems? A walled garden.
What are walled gardens and why are they dangerous to public safety?
The walled garden is a closed system in which one company controls data access, flow and use. They are common in advertising and social media (Think Platforms Facebook, Google, Amazon), but they are increasingly appearing in public safety.
Public safety companies play a key role in modern policing infrastructure, but the unique nature of some of these systems means that they are not always designed to interact in fluidity with the tools of other vendors.
These walled gardens may offer powerful features such as cloud-based bodycam footage and automated license plate readers, but they create an exclusive on how data is stored, accessed and analyzed. Law enforcement is often locked up in long-term contracts with their own systems that don’t talk to each other. result? Fragmentation, sillet insights, and the inability to effectively respond to the community when it matters most.
The people don’t know, so that’s a problem.
Most people don’t understand how much of your personal information flows into these systems. In many cities, your location, vehicle, online activity, and even emotional state can be guessed and tracked through a patchwork of AI-driven tools. These tools can be sold as crime-fighting upgrades, but can be easily misused in the absence of transparency and regulations.
And that is that the data exists not only in a walled ecosystem that is controlled by private companies with minimal surveillance. Tools such as license plate readers, for example, are currently found in thousands of communities across the United States, gathering data and supplying it to their own networks. Police departments often rent out the hardware and don’t own it. This means that data pipelines, analyses, and alerts are determined by the vendor rather than public consensus.
Why should this raise a red flag?
AI needs data to work. However, if the data is locked inside a walled garden, it cannot be cross-referenced, verified or challenged. This means that decisions about who is being pulled, where resources are going, and who is flagged as a threat are made based on partial, sometimes inaccurate information.
risk? Inadequate decisions, potential civil liberty violations, and growing gaps between police departments and the communities they serve. Transparency is eroding. Trust evaporates. New tools are unable to enter the market unless they comply with the constraints of these walled systems, thus limiting innovation.
In scenarios in which a license plate recognition system mistakenly flags a stolen vehicle based on outdated or shared data without reviewing that information between platforms, verifying whether it has audited that decision, or auditing whether the decision has been made, officers can act on false affirmations. We have already seen cases in which flawed technology has led to illegal arrests or escalation of conflict. These results are not hypothetical and occur in communities across the country.
What law enforcement actually needs
Instead of locking data, we need an open ecosystem that supports secure, standardized, interoperable data sharing. That does not mean sacrificing privacy. On the contrary, it is the only way to ensure privacy protections are in place.
Some platforms are working towards this. FirstTwo, for example, offers a real-time situational awareness tool that highlights responsible integration of data that can be published. Others like Forcemetrics focus on combining different datasets such as 911 calls, behavioral health records, and previous incident history to give executives better context on the ground. But what’s important is that these systems are built with public safety needs and respect for the community as a priority, not an afterthought.
Building a privacy-first infrastructure
A privacy-first approach means more than just editing sensitive information. This means restricting access to data unless there are clear and legitimate needs. This means documenting how decisions are made and enabling third-party audits. That means partnering with community stakeholders and civil rights groups to shape policy and implementation. These steps provide greater security and overall legitimacy.
Despite technological advances, we still operate in a legal vacuum. The US lacks comprehensive federal data privacy laws, leaving agencies and vendors to make up for the rules. Europe has the GDPR, providing a roadmap for consent-based data use and accountability. In contrast, the US has a fragmented patchwork of state-level policies that do not adequately address the AI complexity of public systems.
That needs to be changed. Clear and enforceable standards are needed for how law enforcement and public safety organizations collect, store and share data. And you need to include community stakeholders in your conversation. Consent, transparency, and accountability must be burned to every level of the system, from sourcing to implementation to everyday use.
Conclusion: Without interoperability, privacy suffers
With public safety, life is going well. The idea that one vendor can control access to mission-critical data and limit how and when it is used is not merely inefficient. It’s unethical.
We need to move beyond the myth of conflict between innovation and privacy. Responsible AI means a more equitable, effective and accountable system. That means rejecting vendor lock-in, prioritizing interoperability, and demanding strict standards. Because in a democracy, a single company should not control data that determines who gets help, who is stopped, and who is left behind.
