Grammarly Exposed: AI Tool Uses Journalists' Identities Without Consent in 2026

Grammarly Exposed: AI Tool Uses Journalists' Identities Without Consent in 2026

Grammarly is using the identities of real journalists and authors without their permission in its 2026 "Expert Review" AI feature, according to investigations by The Verge and Wired. The tool generates feedback falsely attributed to high-profile figures — including The Verge’s Nilay Patel, David Pierce, and Tom Warren — mimicking their writing style and editorial voice. This raises urgent questions about consent, digital impersonation, and AI ethics in writing tools.

Which Journalists Are Affected?

The Verge’s report confirms that Grammarly’s AI trained on publicly available articles and internal communications to replicate the tone of its own staff. Beyond current employees, Wired uncovered that deceased academics and public intellectuals were also used as training data. Names like Edward Said and Susan Sontag appear in AI-generated feedback, despite no connection to Grammarly or active consent.

How Grammarly Obtains Author Data

While Grammarly’s terms of service allow training on publicly published content, the explicit attribution of feedback to named individuals goes far beyond fair use. The company does not disclose whether it uses private emails, internal memos, or unpublished drafts. Legal experts warn this may violate rights of publicity and misrepresentation laws — especially when users are shown prompts like, "As Nilay Patel would say..."

Grammarly’s Silence and Corporate Growth

Founded in 2009, Grammarly has grown into a $13 billion AI giant, raising $200 million in 2021 and serving clients like HackerOne. Yet the company has issued no public statement addressing these allegations. Its marketing materials vaguely reference "expert writing styles" without clarifying if real people’s identities are being exploited. This lack of transparency erodes trust in AI writing tools.

How to Opt Out of AI Attribution

Currently, Grammarly offers no opt-out mechanism for individuals whose voices are used in its AI training. Users cannot disable attribution to specific names. However, legal advocates recommend filing DMCA takedown notices for unauthorized use of copyrighted material and contacting media outlets to amplify awareness. Privacy-focused browser extensions like Privacy Badger may help limit data leakage, but systemic change requires regulatory pressure.

AI Ethics in Writing Tools: A Broader Crisis

Grammarly’s case is not isolated. As AI tools become more persuasive, companies are increasingly leveraging real human voices — living or dead — to boost credibility without consent. This trend, called "digital voice harvesting," threatens authorship, copyright, and public trust. Without clear regulations around AI training data, we risk normalizing the commodification of identity.

Experts warn this could set a dangerous precedent: if your writing style can be cloned and sold by a tech company, who owns your voice? The answer may lie in new legislation — like the proposed AI Labeling Act — or consumer backlash. For now, users must demand transparency. Grammarly must choose: scale its AI ambitions ethically, or risk becoming a cautionary tale in the age of generative AI.