OpenAI is reaching out to enterprise clients to perform early testing of ‘Skills’ in ChatGPT, bringing closer the prospect of another general-purpose AI platform providing standardised, repeatable legal workflow capabilities.
Skills are reusable task bundles (instructions plus optional files/code) that can be applied automatically and teach AI how to do a specific repeatable task or workflow. They are available in OpenAI’s Codex CLI tool and now also ChatGPT.
Rather than having to write the same instructions over and over again in a prompt, the Skill bundles them together in a way that AI can read and execute. While OpenAI isn’t targeting legal teams specifically, one example might be an in-house legal team that needs to review a non-disclosure agreement (NDA). Teams will be able to include their own detailed instructions on how to conduct the review in a Skill called, for example, Review-NDA.
This follows the recent launch by AI giant Anthropic of a legal plugin for Claude Cowork, bringing with it the ability to execute the likes of automate contract review, NDA reviews and compliance workflows.
OpenAI’s Skills testing is at ‘Alpha’ stage and one person who has been contacted told Legal IT Insider: “While this is not the same thing as Claude Cowork or the Claude legal plug-in, which are higher-level, end-to-end workflow packs, this is a directional signal that the general-purpose AI platforms are all heading towards more standardised, repeatable workflow capabilities over time (including the sorts of repeatable legal/compliance tasks people have been excited about).”
Anthropic’s launch caused the legal software public market to crash and there has subsequently been much debate over the consequences of mainstream legal AI players entering the legal fray.
On the one hand, there should be little surprise that the likes of Anthropic and OpenAI are creating legal plugins (among many other verticals). The familiar arguments around the benefits and drawbacks of big tech v legal tech are being played out here, in much the same way as they have over many, many years.
However, for in-house legal teams that are already leaning towards mainstream AI tools, there is no doubt that these developments are being watched closely.
We have reached out to OpenAI for comment.
The post OpenAI performs early testing of Skills in ChatGPT appeared first on Legal IT Insider.