The ONE Rule to Not Get Screwed at Work with ChatGPT

(ChatGPT, of course, being a synecdoche for the entire landscape of generative AI tools...)

Everybody is trying to figure out how to use generative AI tools. Major corporations are in a panic. Some of them understandably so, especially in light of the recent debacle with Samsung Electronics. Samsung engineers leaked a substantial amount of proprietary information through a large language model, even though it was completely endorsed by company policy to use these tools. This incident is somewhat amusing, a bit tragic, but ultimately an important lesson for us all.

Major corporations are now considering banning these tools or implementing extremely restrictive policies. Employees, recognizing the tremendous benefits of using these tools, will likely find every workaround under the sun. In fact, some are already renaming and rebranding AI tools to bypass lightweight restrictions. I am a contributor to ScalpGPT, or the "GPT API with Built-In Plausible Deniability," a rebranded version of OpenAI's GPT models. It's just wrapper around the official API. We created this in response to a friend whose company banned use of OpenAI's APIs directly.

Part of it was a joke. Part of it was because we wanted to see if we could. Part of it was hoping to get a cease and desist. But mostly it was because we just like doing stuff.

Leaders, of course, have the prerogative to make decisions for their organizations, even if those decisions may be detrimental. If they choose to metaphorically shoot themselves in the foot with a harpoon and remain tethered to the dock while everyone else speeds off into the future on high-speed motorboats, that's their decision. Some employees, however, will do anything within reason to prevent their company from falling behind. They might even drag their leader onto the boat, blindfolded, after having sawed off the chunk of the dock they've harpooned themselves to. It may still be painful, but at least the employees will save the company from itself.

For those employees trying to navigate this challenging environment and avoid being blamed for any negative consequences, there is one guiding principle to always remember: never put anything into a generative AI tool that you would not openly discuss with your fiercest competitor.