7 Key Considerations For Lawyers Negotiating Contracts For AI Services

This post originally appeared on Above the Law. 

Human lawyers will always be crucial for ensuring contracts and licensing agreements related to AI systems and services are comprehensive, clear, and adhere to relevant laws and regulations.

“ChatGPT did a great job reviewing my software contract. It will save me so much on legal fees from now on!” a business professional, John, recently said.

John had run a contract between his company and an AI-powered software vendor through ChatGPT and asked it to pull out the top three areas of concern. He then sent everything to his lawyer for review.

“My lawyer agreed with ChatGPT’s conclusions on two issues. She then explained that the third issue was a judgment call,” John said. “Sure, she added a few things, but I think with ChatGPT, I can handle software contracts myself now.”

“Are you worried about the privacy of your agreement? Or liability issues?” we asked.

“Not as concerned as I am about overspending on legal fees,” John said. 

John’s response highlights a vexing dilemma posed by AI language models like ChatGPT. These tools can display a convincing level of analytical capability that seems indistinguishable from a lawyer’s. But technology-related laws and regulations are highly complex and heavily nuanced. New legal cases and technological discoveries emerge regularly, changing the rules suddenly and unpredictably. 

Human lawyers will always be crucial for ensuring contracts and licensing agreements related to AI systems and services are comprehensive, clear, and adhere to relevant laws and regulations. Negotiations involve many rights and legal issues, including:

Licensing And Usage Rights

A contract for an AI service or system should outline the specific terms of licensing and usage rights granted. Are your rights exclusive? Are there restrictions on sublicensing or the geographic territories where AI services can be used? Include provisions for any limitations on technology modifications and requirements for software updates. 

These rights tie closely with the intellectual property rights your contract may also need to cover.

Intellectual Property Rights

Ownership and protection of IP rights are primary concerns in AI-related agreements. Technology patents, copyrights, trademarks, and trade secrets can be complex and difficult to delineate. Nonetheless, contracts should clearly define each party’s IP rights, as well as any improvements or derivatives stemming from its use. Adequate safeguards should protect those rights throughout the contract’s term.  

Data Privacy And Security

The data sets used to develop AI models can include confidential and sensitive information that requires compliance with privacy regulations like the General Data Protection Regulation (GDPR) in Europe or the California Consumer Privacy Act (CCPA) in the United States. 

Review data privacy clauses to ensure vendors comply with all applicable privacy laws, including data access controls and encryption standards. Clarify which party is responsible for obtaining user consent, managing data breaches, and ensuring appropriate security measures are in place.

Liability For AI Decisions

AI systems can potentially cause harm or losses due to their autonomous decision-making capabilities. Who should be responsible for potential damages or errors—the developer, the user, or the AI system? Addressing liability for harm caused by an AI system’s actions or decisions can include establishing a clear indemnification structure, stating which party is responsible for damages arising from AI-generated decisions, and establishing processes to rectify errors or system failures. 

Increasing transparency into and explainability of AI systems can help mitigate liability risks, which gives you two more contract aspects to consider.

Transparency And Explainability

AI systems may rely on complex algorithms and large datasets that raise concerns about transparency and accountability for their decision-making processes. Consider adding provisions that address transparency requirements like algorithmic disclosures and access to information explaining how AI makes decisions.

Because transparency and explainability can increase under consistent scrutiny, performance metrics are another essential consideration in your contract’s terms.

Performance Metrics

Vendors make powerful claims about their AI system’s capabilities. Can they back them up? Establish performance metrics for success measures. Specify acceptable levels of accuracy and reliability that accurately reflect the real-world performance of the AI system. Make metrics clear, measurable, and attainable to prevent disagreements and avoid disputes later. 

A quality assurance framework outlines each party’s responsibility for monitoring performance, ensuring compliance, and addressing discrepancies. Yet, even the happiest of partnerships can end, so make sure to give termination clauses attention, too.

Termination Clauses

The AI landscape is evolving rapidly. Conditions change with new technological advancements and amendments to regulations. To avoid disputes when your relationship ends, ensure your termination clauses outline processes for handling IP or data rights and establishes knowledge transfer protocols.