Protecting AI Trade Secrets in Litigation: A Congressional Hearing

AI as Critical Infrastructure

• 💡 AI infrastructure is not currently identified as critical infrastructure for cybersecurity purposes, but this should be re-evaluated.
• 🎯 Designating AI as critical infrastructure would allow for examination of minimum cybersecurity requirements and strategic planning with agencies like CISA.

Protecting Trade Secrets in Litigation

• 🔑 The standard process for protecting trade secrets in court involves a protective order issued by a judge.
• ⚖️ Judges balance factors to restrict access to trade secrets, preventing their leakage into the public domain and loss of value.
• ⚠️ The use of third-party litigation funding can increase the risk of trade secret leakage, requiring judges to craft more severe protective orders.
• 🔒 Protective orders can include measures like personal liability for attorneys and strict protocols for information usage, storage, and access logging.

Importance of Trade Secrets in AI Development

• 🚀 Trade secrets are critical to AI development, even for systems described as "open."
• 🧠 Companies do not typically release how they train models, what training data was used, or often the source code, all of which contain significant trade secrets.
• 📈 The sheer size and magnitude of AI systems create an enormous opportunity for building trade secrets, which are essential for maintaining AI leadership and company valuations.
• 🇺🇸 The US leads in AI due to its open system where companies choose what to reveal and what to keep secret, a model that has proven incredibly powerful.