Drupal AI 1.4.0 Release Adds Extensibility Tools, Guardrails, and VBO Support

The Drupal AI module has released version 1.4.0, adding chat processor normalisation, Views Bulk Operations support, developer generation tools, and expanded guardrails for AI requests and responses. Paul Johnson announced the release in a Drupal.org blog post published on 18 June 2026.

The release continues work on the module’s 1.x branch and shifts the project toward a predictable bi-monthly minor release cadence. Paul said future updates are expected to bring new integrations and expanded platform capabilities.

For developers, Drupal AI 1.4.0 introduces markdown editor extensibility, Document Loader integration, AI “skills,” and new drush generate commands. The commands can generate AI providers, AI Automator types and rules, AI guardrails, field widget actions, operation types, AI API explorers, function calls, and function groups.

The release also introduces chat normalisation across processors. The abstraction separates chat interfaces from their underlying AI processors, reducing the need to tie integrations to specific implementations. The update includes a Slack Chat processor and lays the groundwork for a planned AI Agents processor release.

Drupal AI 1.4.0 adds support for running AI Automators through Views Bulk Operations. Site builders can use the integration to run configured AI rules across multiple entities, including workflows for image alt text generation, summarisation, classification, tagging, translation, and editorial automation.

The release expands guardrail controls by allowing them to be configured globally, applied to real-time streaming responses, and used to limit input length. The announcement says input length limits are intended to reduce the risk of “denial-of-wallet” attacks, where large prompts increase provider costs.

The update also lays architectural groundwork for future failover and routing options. The announcement says future support could use tools such as ECA to route AI requests based on provider availability, cost, latency, content type, or custom processing requirements.