In Legaltech News and Law360, Regina Sam Penti and Georgina Jones Suzuki Examine DeepSeek AI Implications for Enterprise Users

The recent release of DeepSeek's generative artificial intelligence (AI) model presents opportunities for the legal technology industry and law firms due to its low cost, but also legal risks – particularly around data privacy and security – given its existing terms of use.

In articles published in Law360 and Legaltech News, IP transactions and technology partner Regina Sam Penti and counsel Georgina Jones Suzuki discussed the implications of DeepSeek’s AI for enterprise users.

Regina and Georgina detail some of the specific legal risks associated with DeepSeek for law firms and other enterprise users, including around data privacy, based on DeepSeek’s online terms of use and privacy policy. While legal risks are particularly heightened with the “chatbot” version of DeepSeek, Regina and Georgina discuss the opportunities that may exist with the open source versions of the models, though advise that law firms and other enterprise users exercise caution in adopting such models.

According to Regina, if law firms and legal tech companies do experiment with DeepSeek's R1 open-source AI model, they should limit their risks by testing in a sandbox environment that doesn't have access to the rest of their network. She also emphasized the importance of training employees around the risks of using DeepSeek models.

Regina and Georgina also further discuss how DeepSeek may revolutionize the use of AI technology in the legal industry. “The significance of DeepSeek is that it's going to usher in a world of democratization of AI, and that is because they've developed essentially a cheaper version of an open-source model that other parties can then further fine-tune and tweak to create more specialized models,” said Georgina. “Over the long term, DeepSeek might enable or hasten this process of creating what we call value-added systems, where you train up the model and it becomes more specialized… And ultimately, this might foster the more ubiquitous, pervasive use of AI more broadly in the industry.”