��������Ƶ

Note: This article is just one of 60+ sections from our full report titled: The 2024 Legal AI Retrospective - Key Lessons from the Past Year. Please download the full report to check any citations.

IP Risk Assessment in Data Flows

Data Collection and Storage:

A 2023 study by the World Intellectual Property Organization (WIPO) found that legal AI systems are increasingly employing automated IP detection algorithms during data ingestion.[54] The provenance of data inputs is important, as this affects the reliability of a given AI output. Blockchain-based provenance tracking is one method of addressing this concern, by creating an immutable record of data while maintaining confidentiality.[55]

Law firms seeking to develop their own legal AI tools will have access to a large volume of client data. A 2024 report by the London School of Economics has forecast that one of the most significant challenges in such firms developing AI tools will be striking the balance between generating these high-quality training datasets using sensitive information and protecting the privacy of individuals.[56]

The potential for the infringement of IP arises during the training of generative AI models on large datasets.[57] This topic has been litigated on over recent years, with notable cases including a US lawsuit by the New York Times alleging that OpenAI and Microsoft trained their models using New York Times articles,[58] and Getty Images in the UK bringing a claim against Stability AI, alleging that their images had been used to train Stability AI's model.[59] Both cases remain ongoing, and will be instrumental in determining how the model training process moves forward.

"There is always a risk of breaching copyright and confidentiality issues that we have to consider with Legal AI. As a business, we almost always work under NDA when engaging with the client. Entering any details of the contract or anything related to the business matter may constitute a breach of confidentiality obligation."

Aleksandra Kozikowska, Solicitor (Construction and Engineering), UK

Output Generation:

Legal AI tools, such as Thomson Reuters' CoCounsel, can increase efficiency in legal research by producing citations for case law with outputs.[60]