Protecting your IP in an AI World

Protect Your Intellectual Property in an AI World

Businesses must carefully balance the benefits of AI use against the need to prevent data leaks stemming from employee use of public AI tools that lack adequate data governance.

A little over a year ago, artificial intelligence (AI) stormed into public consciousness, sparking intense interest across the globe. The release of ChatGPT in November 2022 unleashed a frenzy of activity as OpenAI competitors rushed to bring their products and tools to market. Mass media provoked widespread interest with claims that AI would, at minimum, eliminate millions of jobs and, at its worst, serve as the death knell for humanity. In terms of business disruption, AI was going to be bigger than the internet.

Then, as 2023 drew to a close, the over-excitement began to subside, replaced by a tacit acceptance that GenAI was here to stay and the notion that companies of all sizes would need to adopt it quickly or risk the ability to remain competitive. Of course, many organizations were caught off guard; in fact, few had any idea how GenAI would affect their operations or how to determine that impact. But in the relatively short span of a year, a new line of shadow IT has rapidly emerged.

Almost immediately, everyone had consumer access to LLM-powered AI chatbots. ChatGPT and its close relatives Microsoft® Bing® and Microsoft® Copilot®, as well as mainstream competitors including Google Bard™ and Google Gemini™, were quickly adopted by employees in every functional role— often without appropriate corporate governance. This poses a potentially major problem for companies in terms of protecting their intellectual property. That’s because most freely available GenAI tools utilize user-entered information as a prompt to improve the accuracy of the tool’s responses through ongoing retraining.

What does that mean? Gen AI chatbots use LLMs capable of interpreting and generating language by learning the statistical relationships between words from millions of text documents during a “training” process. As part of the terms of service of many free online GenAI tools, any data passed to the tool during the response prompt can often be used for subsequent ongoing fine-tuning of the LLM to improve its accuracy.

In particular, best practices pertaining to prompt creation encourage users to include sample text to give the LLM an example of what to look for. If the example comes from an internal proprietary document, it has effectively been released into the public domain. As Owen O’Connell, SVP and CIO at Shell said, “We can’t allow things like our corporate strategy to be flowing through ChatGPT."

Dive into the considerations for enterprises looking to adopt powerful chatbots powered by generative AI models, specifically, text-generation Large Language Models (LLMs), using their data environments and explores strategies to alleviate any associated concerns in this FAIR AI Insights article.

Data protection and confidentiality are core aspects of any comprehensive enterprise architecture, but the availability of GenAI SaaS to consumers and the immediate benefits it provides to knowledge workers have increased the risk of data leaks. Organizations must immediately adapt their data governance policies and procedures to manage the adoption and use of GenAI across the organization.

The question then becomes: How can a company use GenAI while protecting its intellectual property?

Essentially, it comes down to adopting an internalized version of a GenAI model. But, as with all cloud services, companies must ask themselves the question: Should we accept the level of data protection offered by the enterprise-oriented GenAI tools that many public cloud providers offer, or should we use private cloud technology?

Michael Carbin, Associate Professor at MIT, said recently, “If you care deeply about a particular problem or you’re going to build a system that is very core for your business, it’s a question of who owns your IP.” When it comes to data privacy for internal enterprise uses of generative AI, the expectations are no different than with other IT applications. A business will want absolute control and oversight over any AI tool using its proprietary data.

One option is a private dedicated IT environment. The other option is public cloud services. The determination as to which option is best for an organization depends on a business's risk tolerance. For highly regulated industries, private environments are most likely the right choice. Public cloud AI may suit businesses that are less reliant on sensitive data.

However, all companies have some sensitive information to safeguard.The bottom line is that GenAI tools can greatly increase employee productivity, as many are already discovering outside of rigorous corporate governance. At the same time, businesses cannot afford to ignore the risk this poses to their proprietary data and intellectual property.

2023 opened the floodgates to unprecedented public interest and adoption of AI and, today, the race is on to implement responsible governance of this potentially transformative yet risky technology.

The way forward is to fully and responsibly integrate GenAI solutions across the business so that employees have the tools they need without having to look to readily available consumer SaaS tools.

 

Join the Conversation: Find Solve on Twitter and LinkedIn, or follow along via RSS.

Stay on top of what's next in technology

Learn about tech trends, innovations and how technologists are working today.

Subscribe
PE CTA

Sign up for an ideation workshop to leverage platform engineering for sustained infrastructure success.

About the Authors

Nic Du Feu

Global Campaign Manager

Nic Du Feu

Nic is a marketer with over 30 years of experience in telecoms and IT based in the UK. Having spent the first 5 years of his career in engineering he then moved through a variety of GTM roles including product management, marketing, business development and consulting which involved traveling throughout EMEA and included a period in North America. During this time, Nic has crossed paths with a myriad of technologies ranging from the physics of optics and wireless through switching and routing to rating and charging applications, all with a focus on understanding the customer challenges that they can solve. Outside of work, Nic and his wife love to spend time with their children’s growing families and maximizing the riot of plants and colors in their garden throughout the year.  

Read more about Nic Du Feu