Last week Vellum released a new version of its publishing software that allowed connections to external large language models (LLMs), a move intended to streamline AI‑assisted writing and formatting.

Within days, a segment of Vellum’s users expressed strong opposition, citing concerns over data privacy, workflow disruption and the platform’s premium‑only positioning of AI tools.

In response to the criticism, Vellum announced that the AI connectivity feature would be removed, restoring the software to its prior state without external LLM integration.

The reversal underscores the sensitivity of the publishing community to rapid AI adoption and highlights the importance of user feedback in shaping product decisions.