In applying STRIDE to an AI endpoint, which activity helps reduce tampering risk?

• A. Mapping data flows to threats
• B. Increasing API throughput
• C. Reducing logging
• D. Removing access controls

Answer: (A)

When you apply STRIDE, you want to uncover where tampering could occur in the flow of data through an AI endpoint. Mapping data flows to threats directly reveals those touchpoints—where data enters, moves, and is processed or stored—so you can identify potential tampering opportunities, such as data inputs being altered or model outputs being manipulated.

This visibility lets you implement targeted mitigations at those points, like input validation, integrity checks, cryptographic signatures, secure communication channels, and tamper-evident logging, all of which help protect data and model integrity.

The other options don’t address tampering risk within this threat-modeling context: increasing API throughput affects performance, not threat identification; reducing logging lowers visibility and hinders detection and auditing; removing access controls increases risk and does the opposite of reducing tampering risk.