AI and Contracts: A New Frontier for Legal Agreements

AI and Contracts: A New Frontier for Legal Agreements

When drafting a contract involving AI, there are several aspects that may differ from a typical software Contract. Understanding these distinct issues is crucial for business professionals and lawyers drafting and negotiating Contracts for software, SaaS, and AI solutions.

Here are some key considerations:

1. Intellectual Property (IP) Rights:

Contracts involving AI should specifically address the ownership and usage rights of AI-generated outputs. It is essential to clarify whether the Customer retains ownership of the outputs or if the Provider claims any IP rights. Additionally, the Contract should outline any licensing arrangements for the use of AI models, datasets, or pre-existing AI technologies.

?

2. Data Usage and Privacy:

Given the reliance on data for training AI models, Contracts involving AI should clearly define how Customer data will be used, stored, and protected. It is important to address data privacy and compliance with applicable laws, such as data protection regulations like GDPR, CCPA, PDPL or any pertinent to your region. The Contract should specify the purpose of data usage, data retention policies, and obligations regarding data security.


3. Confidentiality and Non-Disclosure:

You can’t own data nor can it be patented, and copyright protects only expression, not the information expressed. So if you cannot truly "own" data what sureties do you keep regarding it? Your data could include trade secrets, and ownership terms might help you protect them. Traditional non-disclosure agreement (NDA) terms may not suffice when ML systems reuse Customer data. Providers should be cautious about confidentiality clauses. Customers, on the other hand, should strive to limit the Provider's use of their data and outputs, even if the AI system can reuse them. It is advisable for Customers to treat generative AI outputs as confidential until their sensitivity is determined, potentially including trade secrets. The Customer should add contract terms restricting?use?of prompts and training data as well. The Provider on the other hand should clarify that any Customer ownership does not extend to prompts or other data it?independently?receives or develops.

?

3. Performance and Accuracy:

AI systems, including ML models, are not infallible and can produce errors or inaccuracies. Contracts should address the performance expectations, define acceptable error rates, and outline any service level agreements (SLAs) related to the accuracy or functionality of the AI system. It is crucial to manage expectations regarding the capabilities and limitations of the AI technology. Providers should consider incorporating broad disclaimers to address this inherent risk. Customers, however, should seek assurances regarding inaccuracies.

?

4. Liability and Indemnification:

Contracts should allocate liability between the parties in the event of AI-related issues, such as errors, damages, or legal disputes arising from the use of AI outputs. It is important to consider the potential risks associated with AI-generated content, such as defamation, discrimination, or infringement, and determine the extent of indemnification provided by the Provider. It is important to acknowledge that Providers may have limited control over machine learning outputs. Customers should prioritize IP indemnities concerning the software itself, if not the outputs. Providers, naturally, may resist indemnities due to the inherent uncertainties of ML outputs.

?

5. Ownership and Control of Outputs:

?AI outputs may include the Provider's proprietary data, as well as third-party data, particularly when trained on extensive datasets. Providers should avoid assigning any preexisting rights, such as exceptions to NDAs, and ideally, avoid assignment altogether. An Agreement stating that generating outputs does not create Provider IP, without transferring any ownership, might be sufficient. Customers should strive for IP rights in the outputs but should also recognize the significance of controlling and defining usage rights for the outputs.

?

6. Regulatory Compliance:

Contracts involving AI should address compliance with relevant regulations and industry standards specific to AI applications. This may include requirements related to transparency, explainability, audits, or compliance with sector-specific regulations. Recently EU promulgated its AI Act which is moving forward towards its final form and maybe gets converted into law by the end of this year or early next year and others will follow suit.

?

It is crucial to consult with legal professionals experienced in AI and technology contracts to ensure that the Contract adequately addresses the unique considerations and potential risks associated with AI technology.

?

We at Ascend Legal have a team of experts on legal tech Contracts. We understand the nuances involved and can draft Contracts better suited for the client with this rich understanding base. Do reach out if you have any queries regarding the same.

?

Author is Co-Founder of Ascend Legal and is an expert in technology Contracts and keeps a keen eye on latest developments in this field.

?

https://ascendlegal.co

?

?

Heinan Landa

CEO, Optimal Networks | Author, The Modern Law Firm | 12x Top MSP Globally | CNN, ABC7, FOX5, ABA, CIO, Legal Mgmt

1 年

Lots of value here. Really nice post.

要查看或添加评论,请登录

社区洞察

其他会员也浏览了