Tenable detectors now score threats natively in 10 languages
Original title: How Tenable brought multi-language support to security detectors with agentic translation
A single model replaces translate-then-detect and is about 20–800 times faster per item and an order of magnitude cheaper.
Tenable says Tenable One AI Exposure detectors for AI threats, harmful content, and policy risks were English-only and previously translated other languages into English before detection. An agentic pipeline now transcreates labeled English data into nine more languages, with multi-judge scoring, a refinement loop, and a non-refusing open-source fallback. A fine-tuned multilingual model then detects threats natively in all 10 languages at a one-time cost that does not grow with customer traffic.
Unconfirmed: Speed and cost figures are Tenable’s comparison with its own previous process.
- The prior flow paid for a full translation pass on every non-English request.
- Judges score fluency, meaning integrity, and whether the original label still holds.
- Sensitive items refused by the frontier translator go to a controlled open-source model.
- Naive translation can strip malicious intent or soften a positive into a borderline case.
- Vendors
- Tenable
- Products
- Tenable One AI Exposure
From the original article
6 min read at tenable.com
AI security tools that can only detect threats in English offer limited protection. Here’s how Tenable trained our threat detectors to process nine additional languages and understand thousands more, without the need for a translation model, which increases costs and slows down processing.
By Aviad Moreshet · tenable.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.