Protecting artificial intelligence in the age of DeFi

BY Matthew T. Kitces

What is DeFi?

Decentralized finance (DeFi) leverages blockchain technologies to perform financial functions without the need for traditional centralized institutions, such as banks and exchanges. Through the use of automated programs called smart contracts that are stored and executed on decentralized blockchains, participants are able to engage with one another in nearly any financial transaction that might otherwise have been overseen by a centralized institution. Transaction information is stored and verified using public, decentralized ledgers.

Artificial Intelligence in DeFi

One major benefit of DeFi is the large quantity of public data being generated surrounding financial transactions. This vast collection of financial data can be used to train and improve artificial intelligence (AI) models, such as arbitrage bots that attempt to maximize gain from anticipated asset price fluctuations. Current uses of AI in DeFi technologies only scratch the surface, especially as the amount of available data increases and the types of DeFi services expand in the future.

Protecting AI in the age of DeFi

Proper protection of IP can be crucial to the success of any FinTech entity, including the emerging field of DeFi. While there are many different ways to protect advances in AI, some common approaches involve protecting the underlying data used to train the AI models as a trade secret. For many businesses, the data used to train their AI models can be one of the business's most valuable pieces of IP.

However, when the underlying data is publically available and likely already used by competitors to train their own AI models, entities must employ other techniques to ensure IP protection. For instance, techniques for acquiring or pre-processing blockchain data prior to AI training might be protectable as trade secrets or via patent protection. A company may achieve competitive advantages from how it categorizes, cleans up, segments, adjusts, or otherwise processes publically available blockchain data prior to training its AI models. It is important to identify these techniques and ensure the company's valuable IP is properly protected. Likewise, a company may have protectable IP in the techniques it uses to initially train or optimize its AI models, techniques it uses to apply its trained AI models, and even techniques it uses to compare its AI models' performance to those of competitors.

As technologies advance in the age of DeFi, the reliance on public, decentralized data for training AI models will increase. Especially when competitors have access to the same training data, it is important for businesses leveraging AI to carefully consider and identify their sources of competitive advantage. Even in cases where training data is not protectable, these sources of competitive advantage often involve protectable IP. Once these sources of competitive advantage are identified, protocols can be established and applications can be filed to ensure the business's IP is properly protected.

Nixon Peabody actively tracks these developments in AI and in FinTech/DeFi, and will continue to provide valuable insights and analyses going forward.

author img


Matthew T. Kitces

Department Attorney

Posts By this author