DARPA’s Information Innovation Office seeks information regarding current and emerging techniques and tools for the operational assessment of potential vulnerabilities in DoD-relevant artificial intelligence (AI)-enabled systems. Responses to this RFI will be used to assess the current state-of-the-art and to identify core gaps that may need to be addressed by a future program in AI vulnerability assessment.
DARPA is seeking techniques and tools that:
- consider a spectrum of relevant adversarial access threat models (white box, grey box, black box, hidden box);
- consider not just the AI model, but also vulnerabilities presented by the entire AI-enabled system development and deployment pipeline; and
- consider the platform-specific challenges in operationally assessing vulnerabilities, including environmental conditions, multi-modal sensor ingest, and system purpose.
This RFI seeks information on the current state of art assessment techniques and tools for testing (ie red teaming) AI-enabled battlefield (cyber-physical) systems deployed in tactical, operational, and strategic operating environments. These techniques and tools may include complete pipelines for operational AI red teaming or be pieces that can be utilized to address the areas below.
- AI red teaming framework and autonomous toolkit
- Cyber means of effecting AI-enabled battlefield systems
- Electronic warfare effects for manipulating AI-enabled battlefield systems
- Physical manufacturing of adversarial effects
- Other effects.
Responses to this RFI are due 28 February 2025 via email to DARPA-SN-25-28@darpa.mil.
Document
Request for Information – Techniques and Tools for Vulnerability Assessment of AI-enabled Systems: DARPA-SN-25-28