Artificial Intelligence

A seemingly routine AI clause may determine whether your vendor can train on your data, who owns the output, what happens when an AI output infringes someone else’s rights, and whether you can use the product to perform a federal contract. Standard AI platform agreements push data and liability risk toward the customer, cap general liability at a fraction of fees paid, and condition the intellectual property indemnity on guardrails the customer has to actually be running. The conditions are where coverage gets lost, and most customers have never checked whether they meet them.

Those problems compound as AI moves through the commercial value chain. A model provider’s terms must work with the promises that a software company makes to its customers. A reseller or integrator needs rights it can pass downstream. A government contractor may discover that a commercial AI license can’t deliver the portability or flow-down commitments that a federal solicitation requires. And a company acquiring an AI-enabled business needs to know whether the target owns the technology and has lawful rights to the data used to build it.

The lawyers at Dunlap Bennett & Ludwig help clients identify and resolve these issues before they close the deal.

Our experience in AI commercial transactions and licensing

AI transactions sit at the intersection of intellectual property, commercial contracting, data rights, privacy, cybersecurity, and, increasingly, government procurement. Our AI transactions practice brings those disciplines together. By doing so, we avoid negotiating rights in one agreement that create problems elsewhere in your business. We also know how much that matters to government contractors, because DBL has been a government contractor delivering intellectual property outside counsel services to all of the US Department of Health and Human Services for almost a decade. Further, our co-founder Tom Dunlap co-founded a venture-backed biotechnology company funded through NIH and CDC awards.

Our practical experience extends to our own use of AI. We develop, test, and deploy AI tools internally under confidentiality and governance controls and evaluate AI platforms against confidentiality obligations. As a result, we understand why contractual restrictions that look workable on paper may fail when they reach the engineers and business teams responsible for implementing them. Further, in addition to writing the most recent authoritative book on AI for use in law firm (link to Practical AI for Lawyers on Amazon), Tom Dunlap has taught numerous CLE courses to lawyers in across the country on the use and implementation of AI.

For cross-border transactions, we are the Washington, D.C. metropolitan area member firm of Alliott Global Alliance, an international alliance of independent professional firms where Tom Dunlap serves as the co-chair of AGA’s AI Advisory Board. Where a deal reaches licensing or regulatory questions in another jurisdiction, we can bring in local counsel from legally independent member firms there.

What we do

The terms governing an AI transaction determine what a company owns, what it can sell, how it may use data, which risks it assumes, and whether the agreement works with obligations elsewhere in its business. We help clients structure those terms from initial review and negotiation through closing.

Model, API, and AI licensing
• Draft and negotiate model, API, platform, and other AI licensing agreements
• Draft provisions governing data use and retention, model training, confidentiality, output rights, vendor warranties, indemnification, liability caps, and portability at expiration or termination
• Allocate AI-related rights and risks in customer agreements, including permitted and prohibited uses, output rights, disclosures, warranties, and indemnification
• Advise on what a client actually owns in AI output, including the limits of a vendor’s assignment of output rights where U.S. copyright law requires a human author
• Analyze open-source and model license requirements, including copyleft obligations, license compatibility, and the provenance problem when a model emits code matching licensed material without carrying its notices
• Negotiate vendor terms addressing subprocessors, audit rights, data residency, security, incident notification, and obligations clients must flow through to customers or government agencies

AI provisions in commercial agreements
• Draft AI provisions in master services agreements, SaaS agreements, enterprise procurement agreements, and other technology contracts
• Address training data restrictions, permitted uses, output ownership, disclosure obligations, warranties, prohibited uses, indemnification, and liability allocation
• Draft customer-facing terms that reflect the capabilities and limitations of AI-enabled products and services
• Structure reseller, distributor, and systems integrator agreements involving AI components, including intellectual property rights, sublicense rights, and liability allocation
• Negotiate joint development agreements involving AI technologies, including background and foreground intellectual property, invention ownership, publication rights, and commercialization rights

Training data and content licensing
• Draft training data and content license agreements addressing provenance, permitted uses, audit rights, corpus documentation, sublicensing, exclusivity, and indemnification
• Evaluate whether clients have sufficient rights to use data and content for AI development, including model training and fine-tuning

Federal procurement, data rights, and AI licensing
• Advise federal contractors on Federal Acquisition Regulation (FAR) and Defense Federal Acquisition Regulation Supplement (DFARS) data rights requirements affecting AI models, software, training data, weights, and technical data, including Small Business Innovation Research (SBIR) data rights, now a twenty-year protection period under the DFARS rather than the five years many contractors still assume, and data-rights assertions
• Evaluate whether commercial AI license terms can satisfy government use rights, data portability, documentation, and other federal acquisition requirements, including the AI acquisition requirements of OMB Memorandum M-25-22, which replaced M-24-18 in April 2025, and M-26-04, and the agency-level terms implementing them in the absence of any FAR or DFARS clause governing AI
• Address organizational conflicts of interest under FAR involving AI development, training data, and relationships with government customers
• Counsel on export controls affecting AI, including Export Administration Regulations controls on advanced computing hardware and closed model weights, deemed-export exposure when foreign-national personnel access controlled technology, the treatment of published and open-weight models, and International Traffic in Arms Regulations jurisdiction where AI software or technical data is directly related to a defense article on the U.S. Munitions List
• Advise on rights in AI developed with federal funding, including Bayh-Dole for patentable inventions and the FAR and DFARS data rights clauses that actually govern model weights, training data, and software

AI transaction diligence
• Conduct buy-side and sell-side diligence on AI businesses, technologies, and assets, including model ownership, training data provenance, open-source and model licensing, intellectual property rights, data-rights assertions, and regulatory exposure
• Assess whether a target company’s commercial licenses and vendor agreements can support its customer and federal government obligations
• Address novation requirements under FAR when acquisitions or restructurings involve federal AI contracts and assets
• Counsel on transaction structures involving export controls, foreign ownership considerations, government contracts, and AI technologies