Prompted: 3 Safety Considerations for AI Use

In September, all eyes were on California Gov. Gavin Newsom as the highly anticipated — and quite contentious — AI safety bill sat on his desk. Finally, this week, the result came in: He vetoed the bill. The proposed legislation focused on major safety-related AI concerns and catastrophic harms that could be caused by the innovations of large AI companies, but many were looking to it as the starting point for a new regulatory landscape in the U.S. Now, we're back to the drawing board. 

All the chatter about the California bill may have made you think about AI safety and regulations in our day to day lives. Are there regulations in place right now that affect our AI usage as marketers? And as laws continue to evolve, how can we stay safe regardless of current legislative protections?

A Quick Overview of AI Regulation

The EU remains the only governing body that has successfully passed comprehensive AI regulations. Its AI Act was adopted this summer to much fanfare from the public and protest from AI providers. Tech companies are still making a final push to loosen the requirements set out by the AI Act, which vary based on the risk level of the AI model. For those deemed high risk (for example, models that can impact health and safety, education, employment, or law enforcement), AI providers will need to comply with a robust set of standards that dictate everything from data training to recordkeeping and cybersecurity. The act is not enforceable for at least another two years, so changes and challenges could still be made.

In the U.S., there is no overarching federal legislation regulating the use or creation of AI — yet. Various frameworks and guidelines have been put forth, and more than 120 bills aimed at AI are currently floating around the U.S. Congress, tackling everything from deepfakes and copyright claims to AI-powered nuclear warfare. State laws vary, and we’re still several steps away from any kind of unified regulatory approach.

Staying Safe

While the California bill only dealt with one aspect of regulation, concerns about safe AI use come from all sides. For communicators, data protections, copyright concerns, and responsible AI use are top of mind. This blog is no replacement for your legal team, but following these tips can get you started thinking about your own safety standards when it comes to AI in your communications practice.

1. Protecting Personal Data

One of the top concerns people have when using AI tools is how the tool stores and uses the information you input. Does it keep your data for personalization purposes? Will it train on your data? The answers to these questions vary per tool, so it’s crucial to review the terms and conditions of any tool before use (yes, we mean actually read them instead of just clicking the checkbox). 

Many tools will also provide the option to opt out of training, or to use the tool in some variation of a “private mode” — ChatGPT, for example, allows you to turn off its memory so it does not store information from chat to chat. (You’ll find this in the tool’s settings — not by copying and pasting a chain message on social media.)

Overall, our advice is “better safe than sorry.” Unless you are using a custom-built internal AI tool, avoid inputting any confidential or sensitive information.

2. The Copyright Question

Issues around copyright have plagued generative AI tools since they first entered the market. While your company may need to set its own legal and ethical standards around how AI is used for content creation, we have a few guidelines to keep in mind that you don’t need to be a lawyer to understand. 

In general, you aren’t able to trademark anything AI came up with, so be careful when using AI for branding assets like logos and product names. Different tools have their own restrictions when it comes to commercial use of their outputs, so, again, make a habit of reading those terms and conditions.

Whether human or AI is creating content, copyright law is still the same — what you’re putting out can’t be too similar to something that already exists. Since generative AI is completely based on existing works, it’s especially important to do your due diligence and make sure your content is unique. 

📚 Read more: Consider these three non-content applications for generative AI tools.

3. Setting Your Standards

When it comes to how your brand is using AI, you are the one setting the rules. Defining best practices within your organization ensures employees are all working with the same set of guidelines. As employees continue to have greater access to and understanding of AI, they may go rogue without clear guidance on how these tools should (and should not) be used.

As for your customers, we always advocate for transparency: Don’t do anything you wouldn’t be comfortable telling someone you do. If you don’t want to become a brand known for being misleading or unclear in your AI use, consider creating a public-facing statement of your policies.

AI's Evolution

Here in the U.S., we’re in a bit of a holding pattern when it comes to universal AI regulation. While we’re continually being promised “regulation is coming,” so far only a few case-specific guidelines have emerged. With the future of the AI industry still in flux, business enthusiasm, consumer skepticism, and upcoming legislation will all play their part in shaping AI’s future. As things continue to evolve, it remains necessary for companies to create their own guidelines and strategies for if, how, and when to adopt AI in order to stay safe.

Looking for guidance on your AI communication to customers, clients, or partners? Our communications team is well-versed on AI developments and ready to provide expert guidance for your business.

Prompted, our biweekly blog series, brings you relevant updates and insights on how to strategically add AI-driven solutions into your marketing toolkit. Subscribe to receive updates in your inbox.