Extract from Isha Marathe’s article “E-Discovery Achilles Heel: With APIs, What You See Isn’t Always What You Get”
It shouldn’t be shocking then that generative artificial intelligence, the latest tech innovation, would bring with it its own e-discovery curveball: more application programming interfaces, or APIs.
Since OpenAI rolled out its chatbot, ChatGPT, in November, the company has launched several large language models, offering them to businesses as a web-based public site—and more popularly, as an API for purchase. Legaltech News has reported on several vendors who are using the OpenAI API instead of the public website to avoid data spilling out of their own ecosystem.
To be sure, APIs aren’t new. From Facebook to Slack to Microsoft, tech providers have relied on APIs for myriad integrations.
But the race to leverage generative AI capabilities has further expedited the proliferation of these plug-ins. And while they may be convenient liaisons between applications, they aren’t always predictable when it comes to how the data flows through them, creating issues around the authentication and capture of data. What’s more, with how quickly data types are growing, APIs’ prominencecould set the stage for longer, more complex e-discovery battles.
Micheal Sarlo, the chief innovation officer at Haystack ID, told Legaltech News that the crux of e-discovery issues when collecting data from APIs is that what you see going in isn’t always what comes out.
