Grammarly says it will stop using AI to clone experts without permission

Editor's take

Grammarly is ceasing its AI-powered expert voice cloning feature after Superhuman flagged its unauthorized use of editorial styles. The company stated it made this decision after review, acknowledging that the feature's output was presented as inspired by specific individuals.

This incident highlights the ethical and legal complexities surrounding AI model training data, particularly when it involves the intellectual property of individuals' writing styles. The ability of AI to mimic distinct authorial voices, even without explicit consent, raises concerns for creators and publishers alike, impacting ownership and fair use in the rapidly evolving AI landscape.

Future developments will likely focus on clearer consent mechanisms and robust attribution frameworks for AI-generated content that draws upon existing human work. Further scrutiny will be needed to see if Grammarly implements more transparent data sourcing or if this leads to broader industry standards for AI training on copyrighted material.