AI safety evaluations must account for cascading effects across the algorithmic supply chain
The paper suggests that conformity assessments under the EU AI Act, NIST AI RMF audits, and contractual liability frameworks built on component-level evaluation may leave deployers and integrators exposed to harms that no single audit participant had visibility into — and therefore no one is positioned to catch before deployment.
Researchers Neumann and Singh, presenting at the 2026 IASEAI conference, studied how AI system components interact across supply chains and found that auditing each component separately fails to detect compounding behavioral effects that only appear at the system level.
ⓘ This corpus reflects one attorney's personal review. It is not a comprehensive survey. Verify scope and currency before relying on it for any matter.