Technology companies are in a race to develop generative AI, while other businesses are in a race to deploy it. However, despite all the promises, this new technology is not without its risks. As business leaders invest in new tools, generative AI exposures could cause headaches, losses and even lawsuits.

Business Investment in Generative AI Is Surging

According to polls from Gartner, only 19% of organizations were using generative AI in pilot programs or productions as of March and April 2023. By October 2023, that figure had shot up to 55%. By now, the percentage of businesses investing in generative AI could be even higher.

Businesses are eager to reap the rewards of generative AI, such as improved productivity and lower costs. However, this is a new technology, and many of the potential issues still need to be ironed out. As businesses deploy generative AI tools, they should be aware of the risks.

Risk Number One: Misinformation

Generative AI programs sometimes produce fabricated information, a phenomenon known as “hallucinations.” Some of these hallucinations are obviously incorrect and even amusing. Other instances are far more dangerous. In one example, New York City launched an AI-powered chatbot to help small business owners. According to Quartz, it gave users illegal advice, such as telling bosses they could take some or their workers tips and telling landlords that they could reject tenants based on their source of income.

For businesses, hallucinations are a serious concern. Imagine an AI chatbot that tells customers incorrect information about your services, perhaps promising discounts that don’t exist, or that provides dangerous advice. Problems could also occur if your company uses generative AI outputs to make decisions and those outputs turn out to be false.

Risk Number Two: Data Breaches

If you want a generative AI program to draft replies to emails, summarize documents or analyze data, you need to feed it the emails, documents and data, and that has led to concerns over data security.

According to Ars Technica, people have found a way to read private conversations with generative AI chatbots even though they’re encrypted. This means that malicious actors may be able to access private information without anyone else realizing it’s happening.

With public-facing chatbots, there’s also a concern that users will be able to access information that is supposed to be private. By entering certain prompts, users can sometimes trick AI programs into secret information. In one example, ZDNet says people were able to break ChatGPT and get it to reveal training data by commanding it to repeat the word “poem” endlessly.

Risk Number Three: Bias

Generative AI tools can be vulnerable to bias. When the training data is skewed in favor of one gender, race or other group, the results can also be skewed. If people use the results without understanding how they were developed, bias may run rampant.

Indeed, there is already evidence of bias in generative AI tools. A UNESCO study found that large language models – the models behind generative AI text-based programs – tend to display gender bias, homophobia and racial stereotyping.

Risk Number Four: Copyright Infringement

Most generative AI programs were trained on copyrighted material, and this has led to lawsuits over whether or not this counts as fair use. While the outcome of these lawsuits remains to be seen, users may need to consider whether the material they create using generative AI could infringe on someone else’s copyright. This can happen when a generative AI program produces an image or text that is extremely similar to, or even identical to, someone’s intellectual property that happened to be included in the training data.

This may be a more common problem than advocates of generative AI would like to admit. Copyleaks studied GPT-3.5 outputs and found that nearly 60% of them contained plagiarized content, while The New York Times says AI image generators will produce images of trademarked characters, such as the Joker, Iron Man and SpongeBob SquarePants, when prompted to do so.

While these famous characters would be recognized by the majority of users, a more troubling possibility is that a generative AI tool could create an infringing image that the user does not recognize. If the user publishes the image without realizing that it infringes on someone’s intellectual property, a lawsuit could follow.

Risk Number Five: Worker Disruption

Generative AI has caused workplace disruption and fears that workers will be replaced by people with better AI prompting skills or possibly even by the technology itself.

D2L surveyed workers and found that 52% of Gen Z workers were worried they’d be replaced by someone with better generative AI skills in the next year, while Axios says big tech companies have used AI as a justification for recent layoffs.

Whether or not workers actually lose their jobs, these concerns can have a negative impact on workplace morale.

Risk Number Six: Public Backlash

Not everyone is happy about the meteoric rise of generative AI and its potential impact on workers and creators. Companies using generative AI may draw the ire of consumers. For example, according to Publishers Weekly, Tor Books faced backlash after using AI-generated images in the cover art for one of its books.

Moving Forward with Caution

According to MIT Technology Review, 2024 will likely be the year that AI regulation and policy takes shape. New regulations will provide guidelines for acceptable AI usage and likely create penalties for violations. In the meantime, businesses should be aware of the various generative AI exposures and take steps to ensure that risk management practices and appropriate insurance coverages are in place.

Founded in 1961, Wilson, Washburn & Forster recently joined Alera Group, the nation’s 14th largest independent insurance agency. As part of this national firm, we can provide our clients with even more resources, technical expertise and best practices, while maintaining the local service, claims handling, and community commitment that Florida businesses expect.

Contact us today at 786-454-8384 for a complimentary analysis of your insurance and risk management program by an insurance specialist in this field.