AI Law

Because we build AI systems ourselves, we know how they actually work. That enables us to help companies avoid trouble when building with AI, buying AI, training using data, shipping AI features, and figuring out internal governance.

How We Help

Product Counseling

We help product teams identify and protect against AI risk.

Governance

We help make AI policies reasonable, understandable, and useful.

Training Data Risk

The Courts are unclear as to many aspects of training data liability: we help demystify that process.

Model Output Risk

We help evaluate and protect against output-driven liability for defamation, infringement, and more.

Licensing

We help manage licensing and allocate risk among parties in a rapidly evolving industry.

Representative AI Law Work

  • Helping companies develop useful internal policies for AI use when unauthorized employee uses were discovered
  • Trying to "break" client GenAI models to evaluate the risk of potentially infringing (or offensive) output
  • Handling disputes over an "AI Girlfriend" and the accompanying threat of lawsuit
  • Prototyping an automatic "Romantasy" book generation LLM in view of Federal case law regarding AI training data and fair use

Selected AI Tools

We put our money where our mouth is: some of our best tools are public.

USPTO Examiner Metrics

Review grant rates, rejection patterns, and related USPTO metrics to shape prosecution strategy before you respond.
Unclaimed Subject Matter Tool

Find sentence-level concepts in a U.S. patent family specification that remain unclaimed across related filings.
Claim Amendment History Tool

Compare amendment activity across a U.S. patent family and see how claims changed from filing to issuance.
Antecedent Basis Checking

Flag missing or inconsistent antecedent basis in claim language before it turns into a drafting or prosecution problem.