Nouveauté Cyera : une sécurité plus intelligente pour l'IA et les données qu'elle manipule

Drive URL: coming soon

For a long time, traditional data security played defense. Build walls, write policies, and lock down the places data is “supposed to live”.

Cloud-first work already broke that model and today AI accelerates the gap even further, pushing sensitive information into AI prompts and agent workflows before traditional controls ever see it.

To keep up, security teams must stop chasing data after it moves and start moving with it.

Cyera’s most recent product releases focus on solving five pain points where teams lose time, context, and control:

• Sensitive data entering prompts in web-based AI tools
• Files with sensitive data are multiplying into copies and derivatives by AI
• Security can’t clearly explain risk in business terms, delaying action
• AI systems that are blind to policies before retrieving data
• Privacy operations forced to run on stale inventories

These are connected capabilities designed to give teams continuous visibility and control across the lifecycle of sensitive data.

Browser Shield: See and Protect Browser-Based AI Usage in Real Time

Data exfiltration no longer starts with a misconfigured cloud bucket. It starts with something much simpler: an employee pasting sensitive data into an AI chat window in their browser.

Employees in every department are increasingly adopting AI tools to move faster and become more productive. They paste a paragraph from a contract into an AI chat to “clean up the language.” They drop a customer snippet into an AI tool to “summarize patterns.” They ask an AI companion to “build a full termination plan for a quiet quitter direct report”. 

It feels like typing, but AI prompts are data movement. 

Most organizations still struggle to answer three basic questions: which AI tools are being used, how, and for what purpose. The controls that worked for email and file shares often fall short in the browser, where traffic is encrypted and context on data, user, and intent matters.

That is exactly where the Cyera Browser Shield operates: it provides an inventory of all browser-based AI tools used in the organization and, when an AI prompt is submitted, real-time visibility and policy enforcement to help teams secure AI usage without slowing work down.

It helps teams:

• Discover which browser-based AI tools employees are using (not just the approved ones)
• Map tools to users and subscription type (anonymous, personal, or corporate)
• Detect sensitive data leakage or unethical use attempts in real time
• Enforce policy with clear actions: block, alert, or allow—before prompt is submitted into public models

This is not about trying to stop AI adoption. Blocking alone rarely works. It is about making AI usage visible, governable, and aligned to real risk by connecting data, identity, intent, and organizational context in real-time, ensuring security and compliance are maintained.

Browser Shield is in early availability. Read the blog to learn more. 

Drive URLs:ai_assets_new.png chatgpt_block.png

Data Lineage for Files: Track sensitive files across their entire journey, not isolated events

Most security teams are still piecing together individual logs and endpoint activity to understand how incidents transpire.  

A financial model is created in SharePoint. Someone downloads it to OneDrive to work faster. A revised version gets uploaded to a different folder. A summary is exported and shared externally. Now there are multiple versions across multiple systems, with no audit trail connecting the events.

Traditional activity monitoring can tell you that things happened. It often cannot tell you whether it was the same file, how it propagated, or what the true blast radius is. And in incident response, the first file is rarely the only one that matters.

Cyera’s Data Lineage for Files connects the dots and shows the story of the file, not just individual events. It maps how sensitive files spread as a living graph: who copied what, where it went, and how it multiplied into variants.

Because sensitive files move quickly, File Lineage can help identify likely related “rogue” copies using content similarity, including versions recreated via copy/paste or modified to bypass typical tracking.

Teams use this context to understand exposure quickly and resolve incidents with confidence.

Data Lineage for Files is in beta. Read the blog to learn more.

Drive URL: MainImage.png
Classification Topics: Prioritize what your business actually cares about

Cyera’s new Topics functionality is built for the moment after classification, when teams have plenty of findings but still cannot answer what leadership needs: what matters most, what is urgent, and what gets fixed first.

Topics are an LLM-powered business context layer that uses classifications as signals, then evaluates intent and full-document context to map content to business concepts that teams actually use, like “M&A Planning”, “Pricing Strategy”, or “Customer Contracts”.

With Topics, security teams can:

• Bring their taxonomy to life in Cyera using plain-language definitions, internal terminology, and business-specific concepts
• Define new Topics as priorities change, without rebuilding complex rules
• Prioritize and act on risk in business terms instead of long lists of classifications

For example, a global enterprise found confidential merger planning documents shared via public OneDrive links. Cyera’s Topics surfaced the M&A theme (including an internal codename learned from context) and made the remediation target obvious.

Topics is not keyword matching, and it is not “searching your data and stitching together excerpts” to answer a prompt. It is intent- and context-based classification that turns signals into a decision-ready view of risk.

When your taxonomy matches the business, risk is easier to explain, easier to prioritize, and faster to remediate.

Classification Topics are generally available. Read the blog to learn more. 

Drive URL: Creating Topic.png

Cyera MCP: Build Your Own Data Security Agents

Security teams spend too much time assembling answers or completing workflows using scattered dashboards and manual queries. Cyera MCP changes that. It's a Model Context Protocol server that lets AI tools, such as Claude, Cursor, Microsoft Copilot, ChatGPT, or custom-built agents, query Cyera's data security intelligence using plain language.                                                                                                                        

Instead of pivoting between tools, you ask a question: "Which data stores hold our most sensitive classified data?" or "Where does this individual's personal data appear across our environment?" The answer comes back structured, sourced, and ready to act on.

The use cases go beyond ad hoc queries. Security teams can build agents that continuously hunt for threats or flag and investigate issues such as sensitive datasets with overly broad permissions. SOC teams can automate operations and incident response. Leaders get executive reporting pulled directly from authoritative data with no manual assembly required.

Cyera MCP turns AI from a helpful assistant into an operational security tool, grounded in the data you already trust. 

The MCP Server is generally available. Read the blog to learn more. 

Drive URL: Screenshot 2026-03-27 at 10.58.08 AM.png

Cyera Privacy: Build Your Privacy Program in the AI era

Privacy programs have a problem. Outdated data. Major assumptions. Consent is difficult to track. Data subject requests (DSR) require information from many different sources. Data inventory is a spreadsheet or manual tool that reflects inaccurate processing activities the moment it’s finished. When a DSR request comes in or an assessment is needed, teams chase stakeholders and rely on best guesses.

That gap between privacy operations and the AI reality we now live in is where risk builds.

Cyera Privacy closes that gap by grounding privacy operations with ongoing data discovery and with enriched data classification. Teams operate from an up-to-date understanding of where personal data lives, how it’s used, and where risk exists.

Cyera Privacy offers a holistic solution for: 

• Data Mapping. Keep your records of processing activities (RoPA) accurate as systems, SaaS apps, and AI tools evolve
• Privacy Assessments: Pre-fill assessments with real data and system context, including AI risk assessments. 
• Data Subject Request (DSR) Automation: Streamline the full DSR lifecycle for every request, from intake to audit.
• Centralized Privacy Risk: Consolidate risk across systems, assessments, and AI usage into a single, prioritized view
• Consent Management: Build and deploy code-free consent collection points with automated enforcement across GDPR, CCPA, and more

Drive URL: Privacy screenshot.png

Also we’re enabling Cy for privacy, our AI agent that helps teams generate assessments, surface risks, and answer questions about their data environment using real system context, while keeping humans in control of every action.

Cyera Privacy is now generally available. Read the blog to learn more. 

And more: Make Action Consistent and Auditable  

None of this matters if teams cannot act on it consistently.

This release also includes improvements designed to help teams investigate faster, prioritize with confidence, and drive consistent action with clear accountability and audit trails:

• CNAPP Integration: Bring cloud posture signals together with sensitive data context to highlight misconfigurations tied to real exposure, reduce noise, and align cloud and data teams on what to fix first.
• AI Filters: Use plain-language intent to quickly narrow to the exposures that match a real scenario, improving triage and speeding audit, incident response, and executive asks with targeted, explainable filtering.
• Remediation Portal enhancements: Delegate issue resolution to the people closest to the data while keeping security in control through oversight, approvals, and auditability.
• Action Templates: Standardisez les procédures de remédiation courantes en réponses guidées et reproductibles. Vos équipes pourront ainsi réduire leurs délais d'intervention, garantir des résultats homogènes et passer à l'échelle en transformant chaque interrogation sur la marche à suivre en un véritable manuel opérationnel.
• Assistant de recommandation de politiques DLP: Ne configurez plus vos politiques à l'aveugle. Bénéficiez de recommandations basées sur l'IA qui vous indiquent précisément quoi créer, quoi ajuster et pourquoi. Vos équipes réduisent ainsi des mois de tâtonnements à quelques minutes de revue guidée. 

Tout ce travail n'a de sens que si vos équipes peuvent agir de manière répétable, cohérente et justifiable.

Une sécurité au rythme de vos données et de votre IA

Pendant deux décennies, la sécurité des données a consisté à ériger des murs autour d'emplacements fixes.

Mais les données ne sont pas restées en place.

Désormais, l'IA fait du mouvement la norme à travers les applications, les identités, les agents et les navigateurs. La sécurité ne peut plus reposer sur une cartographie statique de « lieux connus » alors que les données se comportent comme un système dynamique.

Le changement est simple :

• Voyez ce que l'IA voit
• Suivez les données tout au long de leur cycle de vie
• Exprimez les risques en termes métier
• Agissez aussi vite que les données se déplacent

Vos données évoluent déjà au rythme de l'IA.

Il est temps que votre sécurité suive le mouvement.

Si vous souhaitez découvrir comment ces fonctionnalités s'articulent dans votre environnement, demandez une démonstration dès aujourd'hui.

FAQ

Q.) Comment identifier l'IA fantôme et sécuriser les prompts dans le navigateur ?
Cyera Browser Shield offre une visibilité en temps réel et une application des politiques pour les outils d'IA basés sur navigateur. Il détecte les données sensibles dans les invites d'IA au moment de la saisie et applique l'action définie par votre politique (par exemple, avertir ou bloquer), permettant ainsi aux équipes de réguler l'utilisation de l'IA sans dépendre de l'autodiscipline des utilisateurs.

Comment puis-je suivre le moment où un fichier sensible a été copié, téléchargé ou partagé, et évaluer l'ampleur de l'incident ? 
Cyera Data Lineage for Files aide les équipes de sécurité à comprendre la propagation des fichiers sensibles en reliant les copies et variantes associées à travers les environnements. Plutôt que de traiter chaque téléchargement, partage ou transfert comme un événement isolé, File Lineage assure une continuité permettant aux équipes de délimiter l'exposition plus rapidement et de comprendre où les données sensibles se sont propagées.

Comment puis-je hiérarchiser les risques liés à la sécurité des données en termes métier (et non uniquement via des étiquettes techniques comme « PII » ou « PCI ») ?
Topics est une couche de contexte métier basée sur LLM qui utilise l'intention et le contexte complet des documents pour mapper les données classifiées vers des concepts métier (et une taxonomie définie par le client), tels que la planification de fusions-acquisitions ou les contrats clients. Cela aide les équipes à hiérarchiser et à gérer les risques en termes métier plutôt que de gérer de longues listes d'étiquettes techniques..

Qu'est-ce qu'un serveur MCP pour les équipes de sécurité et que puis-je en faire ?
Le serveur MCP de Cyera offre aux équipes de sécurité un moyen standardisé de créer leurs propres agents IA afin d'accélérer les enquêtes et d'automatiser les flux de travail de sécurité. Les analystes peuvent interroger les données de Cyera en langage naturel et obtenir des réponses sur l'emplacement des données sensibles, leur classification et les risques les plus importants. 

Comment puis-je automatiser le traitement des demandes d'accès aux données (DSR/DSAR) et maintenir la précision du cartographie des données et des évaluations de confidentialité à mesure que les données évoluent ?
Cyera Privacy repose sur une découverte continue pour ancrer les opérations de confidentialité dans la réalité actuelle des données. La solution prend en charge la cartographie des données, les évaluations des risques de confidentialité pré-remplies avec le contexte système, des flux de travail rationalisés pour les demandes des personnes concernées (DSR) et une gestion du consentement sans code pour des réglementations telles que le RGPD et le CCPA.

‍