Back to the MIT repository
2. Privacy & Security2 - Post-deployment

IP information in prompt

Copyrighted information or other intellectual property might be included as a part of the prompt that is sent to the model.

Source: MIT AI Risk Repositorymit1296

ENTITY

3 - Other

INTENT

2 - Unintentional

TIMING

2 - Post-deployment

Risk ID

mit1296

Domain lineage

2. Privacy & Security

186 mapped risks

2.1 > Compromise of privacy by leaking or correctly inferring sensitive information

Mitigation strategy

1. Employ Adversarial and Guidance-based Prompt Engineering Strategies, such as Negative Prompting and detailed Task Instruction Prompting, which explicitly instruct the model to steer the generative process away from reproducing identifiable copyrighted or proprietary content. 2. Apply Concept Erasure Techniques to the model weights to localize and eliminate the generative capacity associated with known protected intellectual property concepts, thereby mitigating the risk of inadvertent reproduction. 3. Deploy an Infringement Detection Module within the generation pipeline to identify and flag potentially copyrighted material in the output, coupled with a validation layer that verifies generated content against known IP databases.