Good Legal News for ChatGPT!

Good Legal News for ChatGPT!

For those in need of a little catch-up, ChatGPT stands for Chat Generative Pre-training Transformer and is a chatbot launched by OpenAI in November 2022. OpenAI is funded by Microsoft and several A-list tech gurus, and its use is exploding. It answers questions, creates pictures, writes stories, and can even produce software code. It’s a game changer.?

I recently spoke to a college computer science major who said he’d used it to help write code for a project. He said the output was “pretty good” and just needed a little clean-up. That’s huge. We all know it’s easier to be an editor than a creator. What if this could accelerate code development by 10, 20 or even 30%? That’s a turbo charger for innovation.

But what about the legal ramifications? If employees of companies use ChatGPT to help write code, who owns the output? This is crucial?in software. The traditional boogeyman has been open-source software, also referred to as third party code. Customers and acquirors of software companies have strict requirements around the use of open-source software because they don’t want to be sued for using someone else’s software. So, can you use the output of ChatGPT in your software and not worry about getting sued? We start with the rights you get from the owner of ChatGPT, OpenAI. That terms of use can be found at?https://openai.com/terms/?and say:

????????(a)?Your Content. You may provide input to the Services (“Input”), and receive output generated and returned by the Services based on the Input (“Output”). Input and Output are collectively “Content.” As between the parties and to the extent permitted by applicable law, you own all Input, and subject to your compliance with these Terms, OpenAI hereby assigns to you all its right, title and interest in and to Output. OpenAI may use Content as necessary to provide and maintain the Services, comply with applicable law, and enforce our policies. You are responsible for Content, including for ensuring that it does not violate any applicable law or?these?Terms.

The net-net here is you own what you put into ChatGPT and what comes out as a result of your input. That makes lawyers happy. This is different from some open-source licenses that make you open source the code you produce just because you use their code.?

So are we done? Not quite. OpenAI isn’t leaving it just to terms of use. For the open-source geeks in the room, this is no simple BSD license. The OpenAI terms are like a nesting doll, they refer to other documents you must read as well: Service Terms, Sharing and Publication Policy, Usage Policies, the Privacy Policy, and a description of How Your Data is Used. I call disparate terms cobbled together like this an Easter Egg Hunt, you get to poke around to try and figure out what is really going on. This leads to ambiguity. Turns out ambiguity makes lawyers happy as well.

Let’s start with the Service Terms. They incorporate terms from two other OpenAI offerings, Dall-E for making pictures and Codex for generating software code. The Dall-E terms essentially say don’t do mean things, keep in legal and keep it “G” rated. But the Codex terms have an enormous Easter Egg. They say, “Output generated by code generation features of our Services, including?OpenAI Codex, may be subject to third party licenses, including, without limitation, open source?licenses.” Remember all that about open-source code? Well, back it comes into the picture. But not all is lost. If you do generate code from ChatGPT/Codex, you can run a scanned to find open-source code and look at the licenses to that code to make sure you can use it without being sued.?

The Sharing and Publication Policy is an interesting one and reads a bit like the BSD license mentioned before because it requires attribution for publishers of content like books and short stories.?“The author generated this text in part with GPT-3, OpenAI’s large-scale language-generation model. Upon generating draft language, the author reviewed, edited, and revised the language to their own liking and takes ultimate responsibility for the content of this?publication.”?This isn’t required for software code now, but I can imagine a future time when it is.

The last I will highlight is How Your Data is Used. They say, “we may use the data you provide us to improve our models.” They also say they remove personally identifiable data and only use a small sampling of data per customer. You can opt-out of this by using an organization ID. Seems companies would want to do this. Even with the opt-out, this is a crowd sourced data model learning all the time. Your input becomes part of the collective knowledge. How would this not be part of another’s results at some point? But that is the key to innovation and intellectual property. There is always a balance between the advancement of society and individual ownership of creations.?

This is the fabulous thing about technology law and enforcement, they both lag the technology itself. Fast movers with more risk tolerance will use the technology in new and creative ways while focusing only on the Terms of Use as a license to proceed. More established and conservative companies will limit use of ChatGPT for development purposes until “thoughtful policies” are written considering language outside of the Terms of Use.?

Ultimately, all legal decisions in the corporate world are business decisions balancing risk and reward. So, congratulations! If you’re read this article, you are now better positioned than most to make the right business decision for you or your company about using ChatGPT. It’s all matter of how good this legal news is for you.

Gant Redmon is CEO of Hopara, Inc. , a data viz company in Boston, MA. Visit our homepage.

#openai #chatgpt #datademocratization #dataviz #microsoft #legaltech #venturecapital 微软 OpenAI