Large Language Model (LLM) Security and Integrity Assurance

The contract seeks to establish robust security and integrity measures for large language models used in government applications, with a focus on preventing hallucinations, mitigating prompt injection attacks, and blocking unauthorized access or disclosure of sensitive data. It requires all AI-generated content to be clearly and permanently identifiable as machine-produced to ensure transparency and accountability. The contractor must design and implement technical safeguards that uphold the truthfulness and reliability of outputs, integrating controls that can be independently validated under government supervision. The contract is structured as a subcontract under the NAICS code 541715 and is issued by the Program Support Center Acq Management Service within the Department of Health and Human Services. Proposals must be submitted by July 2, 2026, with performance expected to center in Washington. Independent government testing will be mandated to assess model behavior under adversarial conditions and validate compliance with security and authenticity requirements. The contractor is expected to deliver a secure, verifiable, and auditable AI system that meets federal standards for responsible deployment in public sector operations.