Why Some Cities Have Banned Facial Recognition for Police Use

Imagine walking down a crowded street and being scanned by a camera linked to a police database. Sound like science fiction? It’s not—it’s happening in cities worldwide. But a growing number of places are hitting the brakes on facial recognition tech, especially for law enforcement. Why? It’s not just about privacy concerns—it’s about accuracy, bias, and a fundamental rethinking of what policing should look like in a digital age.

The accuracy problem that won’t go away

Facial recognition systems are not perfect. In controlled tests, top-tier algorithms can identify faces with high accuracy. But in real-world policing—with grainy footage, poor lighting, or partial faces—error rates skyrocket.

• A 2019 NIST study found that many facial recognition algorithms have higher false-positive rates for Black and Asian individuals compared to white individuals.
• The ACLU tested Amazon’s Rekognition software and found it falsely matched 28 members of Congress with mugshot photos—disproportionately affecting lawmakers of color.

For police, a false match can mean harassing an innocent person or even making a wrongful arrest. In Detroit, a Black man named Robert Williams was arrested and detained for 30 hours after facial recognition falsely identified him as a shoplifter. He was standing in his home while the crime took place miles away.

Bias baked into the code

The accuracy gap isn’t random. It stems from how these systems are trained. Most facial recognition AI is trained on datasets that are overwhelmingly white and male. When deployed in diverse communities, the tech struggles:

• A 2018 MIT study showed that commercial facial recognition systems had error rates of up to 35% for darker-skinned women, compared to 0.8% for lighter-skinned men.
• In London, a 2020 police trial using facial recognition at a concert found that 81% of flagged matches were false positives—and Black individuals were disproportionately targeted.

Critics argue that deploying inherently biased technology enforces systemic racism rather than solving it. As one civil rights advocate put it: “You can’t fix a biased system by adding more code.”

Beyond privacy: the chilling effect on public life

Privacy advocates often lead the charge against facial recognition, but the case doesn’t stop there. When police can scan a crowd in real-time, it changes how people behave. This is called the “chilling effect”—the knowledge that you’re being watched can stifle protest, dissent, and even casual conversations.

• In 2020, Boston banned police use of facial recognition, partly because it could be used to track protesters or activists.
• San Francisco went further in 2019, banning all city agencies—including police—from using the tech. The city’s Board of Supervisors argued it created a “dystopian” surveillance state.

Even law enforcement officials sometimes oppose it. Police chiefs in cities like Portland and Oakland cited the lack of independent oversight and the risk of mission creep: today it’s used for finding suspects, tomorrow it could be used for monitoring lawful behavior.

The legal and ethical quagmire

Facial recognition also raises questions about consent. Most people do not actively “consent” to being scanned in public. In some cities, police have tested facial recognition on public camera feeds without any public notification or debate.

• In 2019, the European Union considered a temporary ban on facial recognition in public spaces, citing a “window of opportunity” to get regulations right before the tech becomes ubiquitous.
• Several US states, including Washington and Vermont, have passed laws requiring police to get a warrant before using facial recognition.

The legal landscape is messy. No federal law in the US regulates police use of facial recognition, so cities and states have become the battleground. When tech companies like Amazon and Microsoft began selling this software directly to law enforcement, local governments moved to stop it.

What about the benefits?

Supporters argue facial recognition can solve crimes faster, find missing children, or catch dangerous fugitives. In theory, true. In practice, the trade-offs are rarely balanced.

• A 2021 audit in London found that the police’s facial recognition system failed to correctly identify a single wanted person during one high-profile trial. All 37 matches were false positives.
• In Detroit, the city’s own police department’s internal audit showed that officers used facial recognition in cases where it was not authorized, raising fears about oversight.

When the technology works, it can be a force multiplier. But when it fails—and it often does—it fails on groups already over-policed.

The cities that said no

The list of cities banning or restricting facial recognition for police use is growing:

Outside the US, cities like Brussels and Stockholm have also pushed back. Some European countries, including Germany, require explicit legal justification for use in public.

A tech problem—or a policy problem?

Facial recognition itself isn’t evil. But deploying it in policing without rigorous standards, independent audits, and public consent is a recipe for harm. The cities that banned it aren’t opposing technology—they’re opposing a specific, risky application of it.

The pattern is clear: when you give police a powerful tool with known flaws and bias, the burden falls on those already marginalized. Until the tech can prove it works equally for everyone, and until laws catch up, communities are choosing caution over convenience. That might be the smartest decision of all.