Superhuman has disabled its Grammarly-powered "Expert Review" AI feature after it was found to be cloning the writing styles of real journalists and experts without their consent. The company apologized, acknowledging it "missed the mark" by not obtaining proper permission or providing control to those individuals.
Moving forward, Superhuman will reimagine the feature to give experts explicit choice over whether and how their writing is used for AI training. The company states its broader goal is to integrate AI assistance into workflows while ensuring user agency and transparent improvements.
The main topics covered are the unauthorized use of personal writing styles for AI, the company's response and apology, and new policies for expert consent and feature redesign.
Superhuman says it has disabled Grammarly’s “expert review” AI feature that said its edit suggestions were “inspired by” real writers, including our editor-in-chief and other Verge staff members.
Grammarly says it will stop using AI to clone experts without permission
Now it will ‘reimagine’ the Expert Review feature, and allow experts a choice about participating in future AI plans.
Now it will ‘reimagine’ the Expert Review feature, and allow experts a choice about participating in future AI plans.
“After careful consideration, we have decided to disable Expert Review as we reimagine the feature to make it more useful for users, while giving experts real control over how they want to be represented — or not represented at all,” Ailian Gan, Superhuman’s director of product management, said in a statement to The Verge. “Based on the feedback we’ve received, we clearly missed the mark. We are sorry and will do things differently going forward.”
Yesterday, Superhuman responded by launching an email inbox for writers to opt out of “expert review,” but now acknowledges that it didn’t go far enough. Superhuman CEO Shishir Mehrotra also apologized and commented on the company’s plans in a post on LinkedIn:
After careful consideration, we have decided to disable Expert Review as we reimagine the feature to make it more useful for users, while giving experts real control over how they want to be represented - or not represented at all. We built the agent to help users tap into the insights of thought leaders and experts, and to give experts new ways to share their knowledge and reach new audiences. Based on the feedback we’ve received, we clearly missed the mark. We are sorry and will do things differently going forward.
Our broader vision is to solve the “last mile of AI” by bringing the power of LLMs to people where they are, inside their flow of work, at the exact moment they need it. As we innovate at the edge of AI, it is important to us that we continue to seek the best ways to help people feel agency over technology and that they can shape it to meet their needs.
Thanks for holding us accountable. We are committed to getting it right next time and will be transparent about how we improve from here.
Meridiano
