Gilbert pledges not to sue Anthropic over AI training copies
Daniel Benjamin Gilbert, proceeding pro se in his copyright suit against Anthropic after opting out of the Bartz class settlement, has filed a covenant not to sue Anthropic over “training copies” or “purchased copies” of his book, except copies sourced from piracy, per Chat GPT Is Eating the World. Gilbert argues the covenant moots Anthropic’s fair-use counterclaim, and is separately seeking partial judgment on the pleadings to narrow that counterclaim to piracy-sourced copies only.
Why it matters: The covenant is a pro se litigant's attempt to narrow what Anthropic's fair-use counterclaim can even argue, testing whether a plaintiff can unilaterally moot part of a defendant's case, a question courts have split on in other AI-training suits.
The record: AnthropicAI training & copyright litigation
Via Chat GPT Is Eating the World ↗
Posted to the wire August 28, 2026.