IP Tech Knowledgy

While artificial intelligence can dramatically enhance and accelerate work, incorporating AI-generated material into client deliverables entails significant risk. The very tools that streamline drafting, coding, and workflow design can also create problems ranging from third-party copyright claims to an AI-generated process that ultimately proves unworkable in practice.

AI risks cut both ways. Providers need to protect themselves up front. Customers need to read the fine print and negotiate for the assurances they deserve.

If You are the Provider: Set the Ground Rules

Start by promising only what you can truly control: that your services will be performed professionally, by qualified people, and in compliance with applicable law. AI systems are notoriously probabilistic and prone to “hallucination.” Accordingly, providers should avoid–and even affirmatively disclaim— guaranteeing that AI output will be flawless or immune from claims of infringement.

From the provider’s perspective, agreements should:

• Limit exposure. Offer indemnification only for losses caused by your gross negligence, willful misconduct, or fraud—not every quirk of the AI model;
• Cap liability. For example, state that total damages cannot exceed the fees paid in the past twelve months and exclude consequential or punitive damages; and
• Clarify warranties and disclaimers. Provide a narrow service-level warranty regarding professional work and legal compliance, paired with broad disclaimers that AI-generated output may contain inaccuracies or third-party IP.

Even with strong contract language in place, providers are well advised to set up an internal review process to confirm that AI-generated content is properly licensed and that any AI-driven processes are effective. While contracts can reduce risk, proper diligence can further prevent disputes.

If You are the Customer: Do not Sign Away Your Rights

Has your provider included an AI disclaimer in the contract? Do not simply accept it. If you have bargaining power, push back. Ask the provider to stand behind the deliverable by:

• Warranting that the work, including any AI-generated components, does not infringe anyone’s rights.
• Representing that the deliverable will perform as agreed, including any AI-generated components that may be included.  

While the provider may resist broad open-ended promises, customers can often negotiate a middle ground: a representation that they have taken reasonable steps to vet the AI output or that the provider will indemnify against any infringement claims that arise.

Bottom Line

AI is not going anywhere, and neither are the legal questions it raises. Providers should be transparent about the role AI plays in their work and protect themselves accordingly. Customers should be equally clear-eyed and insist on meaningful assurances regarding any AI components. Get the agreement first, and you will save time, money, and headaches when the inevitable surprises appear later.

Julian Haffner
410-576-4243 • jhaffner@gfrlaw.com