What To Consider When Crafting an AI Policy

What To Consider When Crafting an AI Policy

November 01, 2023

Has your company considered leveraging the power of artificial intelligence (AI)? If so, you’re not alone. Businesses across the world are taking advantage of ChatGPT, DALL-E and other generative AI tools.

You can use these AI tools to generate marketing content, create simple contracts or provisions, and even respond to customer service requests. While there are endless ways to use AI to become more efficient, it’s important to set and enforce parameters for your team. Here are a few key things to keep in mind when drafting your organization’s AI policy:

Understand the privacy policy of the AI tool before permitting use

OpenAI is the creator of ChatGPT. Its current policy states it will not sell user data and will use personal information only to provide and improve services to users. It even includes details on how users can request to see, correct or delete data stored with ChatGPT in compliance with California law. 

While this policy seems user-friendly, do your research and remember that policies can change at any time. Make sure you know the privacy policies of any AI tools you’re considering using at your organization.

Define internal versus external permissible uses

One of the first decisions you’ll need to make is whether you’ll allow AI-generated content to be posted or used externally. While OpenAI represents that responses to user prompts are original, they are based on an aggregation of source data. And the copyright issues are far from settled. 

Additionally, ChatGPT and other similar tools have been known to produce biased or incendiary responses. Sometimes the outputs are just plain wrong or nonsensical; these are known as “hallucinations.”

Needless to say, it’s crucial to review and edit responses carefully, particularly if you allow client- or public-facing uses of AI-generated materials. Even if you’re using AI tools only for internal purposes, you must review the outputs for accuracy and appropriateness. 

Protect your company’s and clients’ sensitive data

While ChatGPT promises not to sell user data, there’s no guarantee that all AI tools will observe the same ethical code. This is particularly important when it comes to your company’s sensitive data and data relating to your clients. Consider whether you’ll allow employees to input client names when using AI tools for market, industry or competitor research.

Provide examples of permissible and impermissible uses of AI

You can clarify your expectations around permissible and impermissible uses of AI by providing examples of allowed prompts. This helps employees visualize the parameters you’ve set. And it lets you share ideas for leveraging AI. 

You might also want to include some do’s and don’ts, like “don’t put in individual names” or “do review information presented as fact in AI-generated content.”

Contact your benefits adviser or legal counsel for guidance

The AI landscape is growing and shifting rapidly, with new use cases uncovered every day. Many small to midsize businesses are seeking outside help with making AI tools work for them. For an expert’s take on the benefits and risks of AI tools, contact your benefits adviser or legal counsel.