Third-Party AI Providers

In certain cases, Tenable purchases AI services to train Tenable-specific models or build Tenable AI solutions. Tenable only uses generative AI in its applications through a platform approved by Tenable’s internal AI Governance Council, which regularly reviews Tenable’s corporate and product use of AI technologies and methods.

Google Cloud’s Vertex AI platform powers the generative AI features described above for Tenable One (see Current Use of AI for Exposure Management). Amazon Web Services (AWS) Bedrock and Anthropic power Tenable Hexa AI, providing foundation models for agentic reasoning, planning, and tool use, along with content safety and policy enforcement. Adversary View, specifically, uses Anthropic’s Claude Mythos model family for its attack-path reasoning.

Tenable’s approved third-party providers access customer data solely to generate a response, under contract terms that prohibit them from retaining that data or using it to train their own models. Data submitted for inference is not logged, cached, or reused by the provider beyond generating that response. The one exception is Adversary View (powered by Anthropic’s Claude Mythos model): Anthropic retains submitted data for 30 days for trust-and-safety purposes, rather than processing it strictly at inference time. Even during this retention window, Anthropic does not use the data to train its models.

Data Retention and Governance

Customer data and inputs or prompts processed in connection with ML or generative AI product features are retained only as long as reasonably required to provide the requested services. The capabilities described above are applied to customer-collected data that resides within the Tenable platform, which is governed by the Tenable Master Services Agreement. More information about how Tenable protects customer data is available at tenable.com/trust/assurance.