Most companies negotiating an AI contract ask an important question: Who owns the output? But that may not be the question that creates the biggest business risk.
A more important question may be: What happens if the AI provider changes the rules?
A company can spend months building a critical process around an AI tool, training employees to use it, building customer workflows around it, integrating its APIs and negotiating data-use provisions with lawyers. Then the provider changes its pricing, limits usage, replaces the underlying model, changes its terms or shuts the product down. The company may have little contractual protection against any of those changes.
Startups face this risk particularly acutely. A young company can become dependent on an AI provider long before anyone realizes that the provider has effectively become part of its critical infrastructure. And the risk is not limited to the provider disappearing. AI products evolve rapidly. A model that performs well today may be retired tomorrow. An API may change. A feature that was free may move to a paid tier. A provider may impose new restrictions on automated use. Its data practices may change.
These changes may be entirely legitimate from the provider’s perspective. The problem for the customer is what happens when a change disrupts a business process that has become dependent on the service.
An AI agreement therefore should not stop with confidentiality, intellectual property and indemnification. It should address change risk directly.
Can the provider materially change the service without notice? If not, how much notice is required? Can it discontinue a model on which the company depends? If so, can the company continue using the existing version for a reasonable transition period? Can the company export its data, configurations and other essential information in a usable format if it needs to move to another provider? Does the customer have a meaningful right to terminate if a material change substantially affects its business?
And perhaps most important: What is the company’s plan B?
That question should be asked before the AI tool becomes too deeply embedded in the business. If switching providers would require rebuilding a critical system, retraining employees and redesigning customer workflows, the company may have created a form of dependence that the contract does not adequately address.
When an AI tool becomes part of a critical business process, portability stops being merely a technical issue. It becomes a business and contractual issue.
The cheapest AI service to license is not necessarily the cheapest one to depend on.
The most important provision in an AI agreement may not be the one describing what the AI can do. It may be the one describing what happens when it changes, becomes unavailable or stops doing what the business needs.
You may also enjoy:
- Evergreen Clause: This is the Clause that Never Ends
- Tariff Increases as Force Majeure Events
- Covenant of Good Faith and Fair Dealing
- Trump’s IRS Immunity Deal Is Not Law. It’s Paperwork.
and, if you like what you’ve read, please subscribe below or in the right hand column.