- The Black Alliance for Just Immigration (BAJI) and UC Irvine Law’s clinics argue that AI-driven U.S. border enforcement violates international human rights laws, specifically the International Convention on the Elimination of All Forms of Racial Discrimination (ICERD).
- Surveillance technologies like autonomous towers and drones mark migrants of color as security threats, leading to increased dangers rather than safety.
- Algorithmic tools like CBP One, the Automated Targeting System (ATS), and ICE predictive models suffer from severe racial bias, lack of transparency, and inadequate oversight.
- Activists propose a "decolonial" approach to AI based on the African framework of Cosmo uBuntu to ensure diverse human representation and accountability in development.
Surveillance and Algorithmic Bias
- Border Surveillance: Systems like Anduril Towers and Small Unmanned Aircraft Systems (sUAS) replace human patrols, forcing migrants into more perilous terrain and resulting in higher death rates.
- Entry Processing: The CBP One app struggled to identify darker skin tones and lacked language support; the Automated Targeting System (ATS) disproportionately flags travelers from certain nations, such as Nigeria, as high-risk.
- Internal Enforcement: ICE utilizes private algorithms like the "Hurricane Score" from B.I. Incorporated and the RAVEn platform to track individuals, often relying on biased local and international law enforcement data.
Application and Adjudication Risks
- Asylum Review: The USCIS Asylum Text Analytics (ATA) tool may unfairly penalize non-English speaking applicants whose narratives resemble others, potentially rejecting legitimate claims.
- Evidence Classification: The AI-powered Evidence Classifier reviews massive amounts of documentation; atypical records can trigger negative outcomes for vulnerable applicants.
Policy Recommendations
- Transparency and Redress: Agencies must notify individuals when AI decisions are made and provide opt-out options.
- Legal Safeguards: New federal laws should prohibit AI use cases that exacerbate structural racism and mandate independent oversight.
- Civic Action: Municipalities are encouraged to refuse information sharing with DHS for AI development purposes until systems are proven non-discriminatory.