Introduction
Grammarly’s latest ‘expert review’ AI has raised an eyebrow‑raising question: who owns the voice you hear? When the company claims to channel subject‑matter experts—some of whom have died—users discovered an unsettling twist: their own manager’s commentary appeared in the feedback. The result? A breach of identity and a broader debate about AI authenticity. In this post we dissect the incident, explore the stakes for writers and privacy advocates, and chart what this means for the future of AI‑powered editing.
The Breaking Point
Grammarly launched the Expert Review feature, marketing it as advice “inspired by” real‑world experts. Wired exposed that the system pulls from a database that includes profiles of deceased professors and even living professionals who have not consented. During a personal test, I saw my boss’s name pop up as an “expert,” with feedback that felt eerily familiar yet was generated by an algorithm. The feature’s lack of a transparent consent mechanism is the core of the issue.
The Stakes
For writers, the promise of expert guidance sounds useful, but when the source is unverified, it can mislead. For companies, the unauthorized use of a manager’s persona risks legal exposure under privacy laws such as GDPR and the UK Data Protection Act. The scandal threatens Grammarly’s credibility—users expect a safe, honest editing tool, not a hollow mimicry of authority.
The Divide
On one side, advocates argue that AI can democratise expert knowledge, offering high‑quality feedback to anyone. On the other, privacy proponents warn that identity‑based content without consent erodes trust. Grammarly’s response—limiting the use of deceased figures and offering a opt‑out—shows the company is trying to bridge the gap, yet sceptics question whether this will suffice.
What It Means
The incident signals a wider shift: AI tools must be transparent about their data sources. Users will likely demand clear provenance for the “expert” voices they receive. For developers, this is a cue to build consent‑based author models and to disclose algorithmic curation to avoid legal and ethical pitfalls.
The Bigger Picture
As AI writing assistants proliferate, the boundary between genuine expertise and synthetic impersonation blurs. Historical precedent—such as early AI chatbots mimicking personalities—highlights the need for regulatory frameworks. The Grammarly scandal underscores that trust is foundational; without it, even the most advanced tools falter.
Conclusion & CTA
Grammarly’s use of unauthorised identities forces us to rethink the authenticity of AI‑generated feedback. The next wave of tools will have to prioritise transparency and consent. How will you feel about an editor that mimics your boss’s tone? Share your perspective at dakik.co.uk/survey.
