Varonis Details Expanded AI Detection and Response Capabilities in Atlas
Summary
Varonis describes expanded Atlas AI detection and response across browsers, endpoints, gateways and agents, using task intent, identity and data sensitivity to detect risky actions and block, quarantine or review them.
Key points
- AIDR is described as monitoring AI use and comparing actions with task intent, identity and data sensitivity, rather than relying only on credentials or known-bad signatures.
- Hypothetical examples show how indirect prompt injection or an overbroad data export could expose information despite valid credentials and approved tools.
- Varonis says Atlas adds AIDR coverage across browsers, endpoints, gateways and agents, with options to alert, block, modify, quarantine or require human approval.
- The MCP Gateway evaluates caller identity, tool, arguments and intent before a tool call; the article says calls are denied if its evaluator is unreachable.
- Atlas records prompts, responses and tool calls on a timeline and can send events to SIEM and SOAR tools.
- The article cites IBM’s 2026 Cost of a Data Breach Report, which it says found 92% of AI-related breaches involved no proper AI access controls.
Article Details
- Event Type
- Product capability overview describing enhanced AI detection and response (AIDR) in Varonis Atlas
- Impact
- The described capabilities monitor AI prompts, responses, identities, and agent tool calls across browser, endpoint, gateway, and agent layers. They are intended to detect risks such as prompt injection, tool misuse, privilege escalation, and sensitive-data leakage, and can alert, block, quarantine, redact, log, or require human approval. The article reports no specific security incident or customer impact.
Vendors
Products
Agent IBACAt the action itself: Agent IBAC compares the instruction the agent received with its reasoning and the tools and data it reaches for, and responds to actions that don’t align with them.Agent ManifestAgents Agent Intent: Agent Manifest and Agent IBAC Detects intent drift, including drift that builds across a session, such as multi-turn jailbreak attempts, and responds in real time.AI GatewayBetween the agent and its models and tools: The AI Gateway’s Runtime Guardrails inspect prompts and responses for manipulation. The LLM Gateway controls which models people and agents can call, and can alert on spendEndpoint ProtectionWhere employees access AI: Varonis Browser Protection and Endpoint Protection evaluate prompts and responses before they’re sent or delivered, and can log, alert, block, quarantine, or redact sensitive content.LLM Gatewayand tools: The AI Gateway’s Runtime Guardrails inspect prompts and responses for manipulation. The LLM Gateway controls which models people and agents can call, and can alert on spend spikes or block a request whenMCP Gatewaycall, and can alert on spend spikes or block a request when a budget or usage policy is exceeded. The MCP Gateway authenticates the caller and evaluates the identity, tool, arguments, and intent before a tool callRuntime GuardrailsBetween the agent and its models and tools: The AI Gateway’s Runtime Guardrails inspect prompts and responses for manipulation. The LLM Gateway controls which models people and agents can call, and can alert on spendVaronis AtlasEnhanced AIDR in Varonis AtlasVaronis Browser ProtectionWhere employees access AI: Varonis Browser Protection and Endpoint Protection evaluate prompts and responses before they’re sent or delivered, and can log, alert, block, quarantine, or redact sensitive content.