DHS' new cloud-based HART system, which will house biometric data on 250M+ people via AWS' GovCloud, promises to significantly expand DHS' biometric operations
Context & Ripple Effects
HART is the database layer underneath a biometric expansion DHS had already begun: its facial recognition system was live at 15 US airports in April 2019 and had flagged 7,000 visa overstayers, with a plan to screen 97% of departures by 2023. Hosting HART on AWS GovCloud is what gives that program room to scale from border checkpoints to a repository covering more than 250 million people.
The scope question surfaced immediately: weeks after this report, DHS filed a proposal to remove exemptions for citizens and green card holders (removing exemptions for US citizens) — then withdrew it after public criticism. HART matters because it is built to outlast any single proposal; the data repository persists even when a specific screening expansion is rolled back.
First-order effects
- DHS gains effectively unlimited storage for biometric records without building its own data centers, and AWS GovCloud becomes the custodian of one of the largest government biometric repositories in existence.
Second-order effects
- A permanent 250M+-person biometric backend lowers the technical barrier for expansions like the citizen-scanning proposal DHS later floated and dropped — the infrastructure is ready whenever the policy reopens, keeping pressure on privacy critics and Congress to constrain scope by rule rather than by capacity.
Third-order effects
- If the pattern holds, federal identity infrastructure consolidates on hyperscaler government clouds: HART at DHS precedes GSA opening Login.gov's IAL2-compliant face matching to all agencies in 2024, pointing toward a government-wide shift from agency-built systems to commercial cloud-hosted biometric platforms.
The trend: US agencies are migrating core biometric and identity systems onto commercial hyperscaler government clouds, making AWS-class providers structural dependencies for state surveillance capacity.