💼 IP Law x AI: What We Learned from a Bay Area Legal Tech Coffee Chat ☕️

Earlier this month, I hosted the second edition of our Legal AI Coffee Chat series in the Bay Area.

This time, our attendees included:

• A former equity partner from a leading Chinese law firm
• Several US-licensed patent attorneys from Big Law and boutique IP firms
• In-house counsel with over a decade of experience in patent portfolio management
• A Legal AI founder building tools for IP lawyers
• And a legal engineer from a major tech company

Our topic? “AI × IP Law” — promises, limitations, and what’s actually happening on the ground.

Here are a few takeaways from the discussion 👇

🧩 1. Data Confidentiality Remains the Biggest Barrier

I started by asking: “What’s the one task in your IP practice you most wish AI could take over?”

A Big Law patent attorney immediately replied: “The real pain is not the task — it’s that we’re not allowed to use AI in the first place!”

Indeed, data security and firm policies came up repeatedly:

• Even with public patent data, litigation teams worry that search behavior might reveal litigation strategy.
• Some firms display in-system warnings against inputting any client-related info into external AI tools.
• A few are exploring internal deployment or use only on pre-approved, IT-controlled platforms.

One firm even partnered with an AI vendor to build a customized version of GPT internally — showing how deep the concern around data leakage and compliance runs.

🤖 2. AI Hallucinations: Still a Real Problem

Lawyers shared mixed experiences using legal AI tools in real practice: