Comment on Securing the ICTS Supply Chain: Unmanned Aircraft Systems

March 4, 2025
Read the Full Comment

The Center for AI Policy (CAIP) appreciates the opportunity to comment on risks associated with foreign adversary involvement in Information and Communications Technology and Services (ICTS) integral to unmanned aircraft systems (UAS). We strongly commend BIS’s efforts to safeguard U.S. national security, critical infrastructure, and supply chains from adversarial manipulation.

As UAS increasingly incorporates advanced software, particularly artificial intelligence (AI) for navigational assistance, sensor data processing, and limited autonomous operations, they present new vulnerabilities for exploitation by foreign adversaries. Current supply chain integrity requirements inadequately address critical AI-specific risks, including dataset poisoning—where adversaries introduce corrupted training data—and embedded backdoor vulnerabilities, enabling covert remote control or disruption of drone operations.

To close these vulnerabilities and protect against adversarial interference, we recommend BIS adopt three essential measures:

  1. Enhanced AI Supply Chain Security: Mandate rigorous provenance verification for AI training datasets, implement auditable records for all third-party AI model contributions, and require strong cryptographic authentication protocols for AI software updates.
  2. Mandatory Human Override Mechanisms: Establish requirements ensuring that UAS operators can swiftly and securely override or disable AI functionalities during unexpected or erratic drone behavior, thus preventing substantial harm.
  3. Comprehensive Robustness Certification: Require thorough stress-testing protocols and certification processes to validate UAS resilience against adversarial tactics, corrupted sensor data, GPS spoofing, and unforeseen environmental anomalies.

Implementing these AI-specific safeguards will substantially reduce the risk of adversarial interference, enhance drone reliability, and reinforce the security and safety of U.S. persons and critical infrastructure.

Read the full comment here.

Comment on AISI's Second Draft: Managing Misuse Risk for Dual-Use Foundation Models

CAIP's feedback includes five proposed changes to strengthen the final document.

Read more

CAIP Letter to OMB Supports AI Testing in Government Procurement

CAIP shared suggestions regarding the implementation of President Trump's Executive Order 14179.

Read more

Response to OSTP RFI: Items to Include in the Trump 2025 AI Action Plan

Bold national leadership can steer AI in a direction that promotes human flourishing.

Read more