This Week in Startups AI Copyright & Training Data w/ Chris Paniewski | Wilson Sonsini Startup Legal Basics
22 snips
Sep 11, 2025 Chris Paniewski, a partner at Wilson Sonsini, shares his expertise on AI copyright law, particularly its impact on training data. He explains why current laws are unsettled and the legal distinctions between scraping and licensing data. The discussion dives into the concept of 'fair use,' clarifying its role as a defense rather than explicit permission. Paniewski emphasizes how understanding these nuances is essential for startups and investors alike, especially as they navigate their legal responsibilities in the dynamic AI landscape.
AI Snips
Chapters
Books
Transcript
Episode notes
Jurisdiction Shapes AI Copyright Risk
- U.S. copyright rules for AI training vary greatly from other countries and change by jurisdiction.
- Founders must plan where they train, launch, and source data because legal risk differs globally.
Prefer Licenses Over Dicey Defenses
- Get licenses where possible because a signed agreement removes legal uncertainty around training data use.
- Relying on unclear defenses like fair use increases risk for commercial products and exits.
Jason's Personal Licensing Payout
- Jason shared he received $2,500 from Microsoft via his publisher for use of his content in training deals.
- This illustrates that some creators negotiate direct payments for dataset use.
