Metomic Screens Claude Enterprise Prompts for Sensitive Data Before the Model Sees Them

LONDON - August 12, 2026 - Metomic today added support for Anthropic’s Claude inference hooks to its AI data security platform. Claude Enterprise customers can now apply their own sensitive data policies to prompts sent to Claude, and Metomic returns a decision on each request before inference.

Anthropic released inference hooks in beta on August 5. Organizations route employee prompts through a security service of their choice, which returns an allow or deny verdict before the model runs. The hook waits up to five seconds for that answer, and it covers claude.ai, Claude Code, and Claude Cowork. Metomic now works as the inspection layer behind it.

Anthropic built the routing. Deciding what counts as sensitive stays with the customer, and that is the layer Metomic supplies. The same engine that classifies data across Slack, Google Workspace, Salesforce, and more now scans prompts heading into Claude.

“Anthropic has put a policy checkpoint in front of each enterprise prompt and handed the decision to the customer, which is the right call,” said Bethany Ayers, CEO of Metomic. “But the gate is only as good as the intelligence behind it. Wiring a keyword list into a hook gives you a fast filter, not a good decision. It blocks legitimate work, teaches people to route around security, and leaves you with a control nobody trusts. Knowing what counts as sensitive inside messy enterprise data is the hard part. That is the work we have spent years on, and it is what now sits behind the hook.”

Metomic needs no new infrastructure and no agent software on employee devices, and teams can have the integration live in under 30 minutes. They start in observation mode, where Metomic records and analyzes traffic without blocking anything, then switch to enforcement once they have a baseline.

Claude Enterprise customers can use Metomic’s support for Claude inference hooks today. Custom sensitive data labels, which let security and compliance teams define sensitive context in natural language rather than keyword lists, are in early access.

Details of the integration: https://www.metomic.io/solution/claude-inference-hooks/

About Metomic

Metomic is an AI data security platform that gives security teams visibility and control over sensitive data across SaaS applications and generative AI tools. Metomic combines semantic detection with automated remediation and coaching for employees in the moment, so organizations can adopt AI without slowing their people down. Metomic is SOC 2 Type II certified and works with security teams at Zappi, Juni, Zoopla, EcoVadis, Wrapbook, Oyster, and Codat.

Media Contact

Jessica Hansmann, Marketing Lead, Metomic press@metomic.io