CBP signs Clearview AI deal to use face recognition for 'tactical targeting'

Government Use of Private Surveillance to Skirt Limits

• Many argue this deal exemplifies a broader pattern: governments outsourcing constitutionally dubious surveillance to private firms, then buying the data back (e.g., data brokers, telcos, banks, Clearview).
• Strong claim: it makes no sense to ban government surveillance if private entities can legally collect the same data and sell it back; the collection itself must be regulated or banned.
• Counterpoint: in the Clearview/CBP case, no law is being “avoided”; facial recognition and bank-data access are already legal under existing statutes and Supreme Court rulings (e.g., Bank Secrecy Act challenges failed).

Fourth Amendment, Third-Party Doctrine, and Privacy Expectations

• Debate over whether scraping public social media or cloud data violates the Fourth Amendment:
  • One side: no reasonable expectation of privacy in public posts; third-party doctrine allows most searches of cloud data.
  • Other side: expectation of privacy is itself a legal test shaped by social norms; Carpenter v. United States shows that long-term digital records can be protected despite being held by third parties.
• Some call the third-party doctrine “invented” and overbroad, arguing it should be narrowed or abandoned, and that the government shouldn’t be allowed to buy data it couldn’t constitutionally collect.

Policy Proposals and Regulatory Ideas

• Ideas floated:
  • Ban or sharply limit private facial recognition and surveillance camera analytics (local-only storage, short retention, no AI training/marketing, no “consent by entry,” private right of action).
  • Bar the government from accessing data (via purchase or subpoena) that would be unconstitutional to collect directly.
  • GDPR-style rules to impose real costs on data collection and require annual disclosures of all data held, with per-item penalties.
  • A proposed constitutional amendment guaranteeing strong anonymity in finance, travel, and daily life, with deanonymization only after a documented crime—widely viewed as politically unrealistic.

Technology, Inevitability, and Countermeasures

• Discussion of how Clearview differs mainly by linking embeddings to a massive identity database; underlying tech is similar to other facial-recognition systems.
• Historical note: facial recognition has long relied on “AI”/ML; deep learning made it robust across conditions but biased training data remains a concern.
• Skepticism that masks or minor obfuscation will work long-term given face, gait, and device tracking; some argue the “battle for anonymity” may be structurally losing.

Ethics of Building Surveillance Tools

• Several comments stress moral responsibility of engineers: working on tools used for mass surveillance or targeting is seen as enabling authoritarian practices.
• Others note that many “respectable” tech employers (major cloud and social platforms) also facilitate state surveillance or social harm, blurring lines between “good” and “bad” companies.