Brian Brackeen, CEO of facial recognition firm Kairos, on why the tech, given its abuse potential and bias against people of color, shouldn't be used by police
Context & Ripple Effects
Brian Brackeen's argument is notable because it comes from inside the industry: Kairos sells facial recognition, yet its CEO is telling police not to buy it, citing racial bias and abuse potential. Two months later, The Verge convened experts on whether it is time to regulate the technology's police use — the same bias concern Brackeen raised, now a mainstream policy question.
The arc since then has run the other way from restraint: by late 2021 Clearview had scraped over 10 billion web images for police search tools, and a 2025 Washington Post investigation found 15 departments across 12 states arresting suspects identified solely by facial recognition. Brackeen's 2018 warning reads today as an early insider dissent against exactly that trajectory.
First-order effects
- Police departments evaluating vendors face a credibility problem: one supplier is publicly arguing the product should not be sold to them at all, forcing procurement conversations to address bias and evidence standards up front.
- Kairos cedes the law-enforcement segment to rivals willing to serve it — most aggressively Clearview — betting its commercial future on non-policing customers instead.
Second-order effects
- The vendor market splits into two camps — police-facing scrapers like Clearview versus ethics-constrained sellers like Kairos — letting buyers and regulators treat 'facial recognition company' as no longer a single category.
- Civil-society pressure shifts from attacking the technology in the abstract to demanding transparency about which departments use which systems, the gap Vox's reporting identified between deployment and oversight.
Third-order effects
- If the pattern holds, self-restraint by individual CEOs proves insufficient and the binding constraint moves to government: regulation of accuracy standards, evidence thresholds for arrest, and audit rights becomes the decisive lever over how the technology is deployed in policing.
The trend: Facial recognition in US policing is drifting from a vendor-ethics debate toward a regulatory fight, as documented misuse outpaces voluntary restraint.