Intellectual property includes valuable creations and proprietary knowledge, such as an algorithm. Treating data as only customer records can miss sensitive business assets: the algorithm itself may be commercially important and worth protecting. section 8.6.1
Using generative AI therefore requires two practical checks: decide whether the material may be sensitive before submitting it, and examine ownership and licensing before using the output. Copyright status can vary by jurisdiction and use case, while third-party inputs may require appropriate licenses. section 8.6.1 section 8.7.2
Before using proprietary material or model-generated content, check:
- Whether the information includes confidential business knowledge, intellectual property, or an algorithm that competitors could value. section 8.6.1
- Whether anonymization, prior treatment, or avoiding submission altogether is appropriate. section 8.6.1
- Whether the AI provider’s terms, third-party licenses, fair-use rules, copyright, and trademark implications permit the intended use. section 8.7.2
- Whether legal counsel should review commercial use or any situation where intellectual property rights may be at stake. section 8.7.2
The algorithm is the sensitive asset
An analyst shared no usual training data with a model and assumed sensitivity rules were satisfied. The algorithm itself was proprietary and potentially valuable to competitors, so submitting it still created a risk. section 8.6.1
