Control Access to AI Models and Data in Production - AI Mitigation

Adversaries can use unrestricted API access to gain information about a production system, stage attacks, and introduce malicious data to the system.

Access controls on model APIs can deny adversaries the access required for black-box optimization methods.

Access controls can limit API access and prevent cost harvesting.

Access controls on model APIs can restricts an adversary's access required to generate adversarial data.

Access controls on models APIs can reduce an adversary's ability to produce an accurate proxy model.

Access controls on model APIs can prevent an adversary from excessively querying and disabling the system.

Controlling access to the model in production can help prevent adversaries from inferring information from the model outputs.

Adversaries can use unrestricted API access to build a proxy training dataset and reveal private information.

Use access controls in production to prevent adversaries from injecting malicious prompts.

Authentication on production models can help prevent anonymous chaff data spam.

Use access controls in production to prevent adversary's ability to verify attack efficacy.