OpenAI is testing a Writing Style feature that learns a user’s voice from emails, messages, and files to apply a personalized tone, wording, and formatting to future drafts.

Currently in a limited beta, onboarding notes indicate the AI will write in the user’s voice by drawing from examples in connected apps.

To enable it, users must connect ChatGPT Work to daily tools like Gmail, Google Drive, Slack, and SharePoint so the system can analyze their writing samples.

questions remain about data retention, training data usage, data sovereignty, and how long samples are stored on OpenAI’s servers.

The article emphasizes that this is an initial testing phase with practical limitations, not a guaranteed broad rollout yet.

The potential benefits include more natural, professional tone and consistent messaging, weighed against privacy and data-access concerns.

Additional benefits cited are more natural-sounding automated replies and consistent style across communications.

The piece notes broader AI-safety concerns, echoing past examples where AI guidance led to problematic outcomes, underscoring caution with AI advancements.

OpenAI hasn’t released detailed technical specifics, but the method aligns with established natural language processing practices.

Risks include potential displacement of writers, misuse for impersonation, and trust/privacy issues from cloning a person’s writing style.

Source of information: Digital Trends.

If rolled out, safeguards such as watermarking AI-written content or date-locking initial samples are discussed, though these may be vulnerable to changes.