Compliance Corner is a new series that provides information and guidance on legal and compliance matters within the workplace. This series, a collaboration across several JMP departments, will cover various topics and relevant legal issues. The series aims to help employees understand their responsibilities and avoid legal pitfalls.
In this inaugural installment of our new series, we'll dig into the legal implications of artificial intelligence (AI) within the workplace.
As artificial intelligence (AI) technology advances, its integration into business processes is increasingly common. AI can assist with many business use cases, including advertisements, pricing decisions, fraud detection, recruiting and onboarding, and customer service chatbots, to name a few. However, employees who deploy AI without adhering to SAS' Generative Artificial Intelligence Policy and Generative Artificial Intelligence Policy FAQ can expose JMP to significant legal risks.
Top three legal risks
- Privacy violations: AI systems can process vast amounts of data, including personal and sensitive information. Employees who deploy AI without following company policies may inadvertently breach privacy protection laws.
- Example: You may not ask Midjourney to generate an image of any specifically identified person, such as a famous athlete or politician, or upload a photo of a colleague to create an AI avatar, without the person’s permission (even if that image is already on the internet). That use may violate that person’s privacy rights or rights of publicity.
Intellectual property (IP) infringement: All employees are responsible for reviewing and editing drafts of written materials to ensure they are accurate, align with JMP messaging, and do not include content that could be considered a breach of IP. As a general rule, JMP employees may not use any nonpublic, unlicensed, or private materials in a prompt to a Generative AI system.
- Example: You may not paste a newspaper article of a competitor’s product marketing into a ChatGPT prompt. That material is likely subject to restrictions even though it is online. The content is copyrighted, and you do not have the right to license it to OpenAI, for example, for its use in further training its models.
- Example: You may not paste JMP documentation into a ChatGPT prompt. JMP owns the copyright in its documentation, and JMP has not licensed that content to OpenAI, for example, for its use in further training its models.
Liability for errors and misuse: Generative AI on its own is not a reliable source. If deployed improperly, it may cause errors or be misused in ways that harm customers or stakeholders. You may use generative AI during the initial stages of research or to supplement other research methods.
- Example: A sales representative uses ChatGPT to ask for the biggest names within the industry, but they should double-check that (1) each company in the response is a real company, and (2) the sizes, presence, and offerings of companies are accurate. Additionally, the sales representative should complete other research to ensure they do not miss important competitors.
However, please note there are additional external use cases that require further Legal and manager review:
- Generating software source code.
- Writing software documentation.
- Preparing marketing or sales collateral.
- Drafting communications with any external party.
- Generating any imagery for external use.
By following established guidelines, JMP can harness the benefits of AI while safeguarding against legal and operational pitfalls. To understand generative AI more generally, check out our previous post Ethical innovation with Generative AI - JMP User Community.
If you have JMP-specific questions about the use of AI, please contact Sheila Spence (@SheilaSpence).
If you haven't already, please be sure to complete the Regulating AI training that is required for all employees.
