Forescout Medical Device Security

Artificial intelligence is a layer on a discovery and enforcement platform that works without it. Forescout's medical device security, formerly CyberMDX and now sold on the Forescout Vistaro platform, discovers medical devices with "a wide range of discovery techniques" and "cloud-powered classification", groups them into a taxonomy, and detects threats with "a two-stage threat detection engine that uses a blend of five techniques".

The learned component is VistaroAI, described as "Skills Based Agentic AI + Generative AI + Threat Contextualization", and Forescout's May 2024 AI release refers to its model and prompts. Discovery, classification, FDA recall matching, segmentation and access control carry value without the generative layer.

The product page presents complete discovery as the result, and enforcement has no published approval step. It offers "asset intelligence for every connected medical device, with high-fidelity, cloud-powered classification" and quotes a director of biomedical engineering: "Forescout's solution automatically identified all connected medical devices on our network including model numbers and MAC addresses".

A buyer who takes that at its word has no reason to reconcile the inventory against the CMMS, which is the step that would catch a device the platform missed or misclassified. The platform also classifies and groups devices automatically and enforces network segmentation and access control on them. Nothing on the product page or the AI release describes who approves a segmentation or access change on a device in clinical use, or what happens after a wrong one.

The architecture is described in general terms and nothing is identified. The product page describes many discovery techniques, cloud-powered classification into a medical device taxonomy, and a two-stage threat detection engine blending five techniques, without naming the techniques; VistaroAI is described as agentic and generative AI with threat contextualization.

Forescout's AI release says it uses "secure prompt engineering to control the data available to the model" without naming the model or its provider. No model, version, training data description or accuracy figure is published.

The architecture is set out, with no model provider named. Classification is "cloud-powered", and VistaroAI uses a generative model whose provider Forescout does not name; its May 2024 AI release says customer data does not train the model and prompts are not embedded. Forescout's privacy policy points customers to a data processing addendum and product privacy datasheets for how product data is handled, and the trust portal may say more. No subprocessor list or cloud host appears on the product page. The product's lineage is traceable: it was CyberMDX, acquired by Forescout in 2022.

Scale figures, an award and one quote, with no measured outcome. The product page reports more than 10 million devices secured, 42,000 facilities, 417,000 hospital beds and more than 300 healthcare providers, and quotes Vince Rosati, a director of biomedical engineering whose organization is not named. Frost & Sullivan gave Forescout its 2025 Global Medical Device Security Product Leadership Award, an analyst recognition rather than a test result.

Forescout's Vedere Labs publishes vulnerability research on connected medical devices, which describes the problem rather than the product's performance. No classification accuracy, detection rate or false positive figure is published.

Categorical commitments on AI are published; retention and de-identification are not. Forescout's May 2024 AI release says "Forescout does not train its model using customer data", that "prompts are not embedded", and that secure prompt engineering controls what data the model sees. The platform holds a continuously updated map of every connected device on the clinical network, with configuration history.

Its privacy policy excludes customer data processed through the products and points to a data processing addendum and product privacy datasheets for the terms that apply; no retention period is published on the product page, the AI release or the privacy policy. Forescout says it reduces the risk of hallucination and bias through feedback mechanisms and regular audits, without describing guardrails or how a safety event is handled.

A data processing document reaches the service; no business associate position is taken. Forescout's privacy policy (effective 15 August 2024, updated 1 March 2026) covers its websites and portals and says it does not apply to personal information processed for business customers, directing subscribers to a data processing addendum and product privacy datasheets. No page names HIPAA, protected health information or a business associate agreement, although the product observes clinical network traffic.

A recognized authorization is named in the company's own material, without the report. Forescout announced on 24 March 2026 that its cloud services received a FedRAMP High authority to operate after an independent assessment against the High baseline, and it publishes material on its Common Criteria certification, which tests product security features. No SOC 2 or ISO certificate is named in its public material, which leaves the TrustShare trust portal as the likely place for one.

No device claim; the product works the device regime for the customer's fleet without stating its own position. Forescout's medical device security is network security software with no FDA clearance and none claimed. The product page offers "insights into FDA class and recall status to help ensure security without impacting patient care" alongside cyber risk for each device. No page states an intended purpose, or says how segmentation of a device in clinical use fits the hospital's device risk management.

Responsible AI is stated as practice, with no evaluation behind it. Forescout's May 2024 AI release says it "lessens the risk of hallucination and bias by leveraging feedback mechanisms and conducting regular audits" and uses secure design principles. No framework, audit result, model evaluation or detection accuracy is published.

The failure that matters is a device misclassified or a legitimate behavior read as a threat, followed by segmentation or access changes on a device in clinical use; no result is published by device type, manufacturer or protocol.

Mechanisms let a team examine what the platform concluded; nothing stands behind it. The product keeps a "Searchable history of asset configuration changes over time" and its threat detection presents correlated, enriched and contextualized data, so a security or biomedical team can trace an alert or classification. No accuracy or false positive figure, warranty, remediation commitment or correction route for a misclassified device is published on the product page or in the AI release.

It has no clinical record surface; integration with the hospital's other systems is claimed, with none described. The product page lists ServiceNow as a strategic alliance partner and the Forescout bundles include ecosystem integration modules, but the medical device page names no CMMS, ticketing, SIEM or record system connection and describes none in either direction. The platform itself enforces network access control and segmentation.

Deployment options are stated, including ones where data stays on site; cloud residency and isolation are not. The product page lists on-premises installation, virtual machines and Docker containers, air-gapped systems, Forescout appliances, hybrid configurations, fully cloud-based operation, standalone sensor appliances and installation on routers and switches. An air-gapped or on-premises deployment keeps device data inside the hospital. For the cloud services, which hold a FedRAMP High authorization, no region, residency commitment or tenant separation is stated, and classification is described as cloud-powered.

No price, and the packaging is published. Forescout sells the platform in four bundles, Cyber Exposure, Security Core, Security Automation and Cyber Assurance, listing which capabilities each includes (access control, asset intelligence covering medical devices, segmentation, cloud management, VistaroAI, threat detection and response); there is no separate healthcare bundle. No price, tier figure or licensing unit is published, and deployments can involve Forescout appliances.

Coverage is named, with validation behind part of it. The product covers medical devices alongside IT, OT and IoT across the hospital network, for security and biomedical engineering teams, and Forescout reports more than 300 healthcare providers, 42,000 facilities and 417,000 hospital beds. The one quote comes from a director of biomedical engineering whose organization is not named, and no hospital is named as a customer on the product page.