eSentire buys stealth AI security startup, ships Atlas AIDR

eSentire announced it acquired a stealth-mode AI security startup and fully incorporated the product into its Atlas Platform as Atlas AIDR. The module is meant to give security teams a real-time record of AI activity — conversations, tool calls, tokens/cost, and security events — with inline controls for PII redaction, prompt-injection screening, and agent/MCP tool governance.

SAFETY desk — same-day primary acquisition shipping into a live MDR platform shows enterprise demand for audit trails and inline guardrails as employees and agents use AI at work.

What AIDR is meant to do, still company: give security teams a complete real-time record of AI activity across their estate — the full conversation timeline, every tool call with its inputs and outputs, token and cost accounting, latency, and any security events — all attributed to a user, team, and workstation.

Platform, as the same page has it: Atlas is vendor-agnostic, and AIDR continues that — it captures and governs AI activity and complements the endpoint, SIEM, and other security tools customers already have. SIEM here means security information and event management — the log-and-alert system a security team already runs.

Capabilities listed inside Atlas Detect, still company: complete AI visibility — users, teams, and workstations tied to sessions, plus a live inventory of agents, models, providers, MCP servers, and tools; real-time security and inline control — redact personally identifiable information (PII) before it reaches the model, screen for prompt injections, and enforce custom business-logic rules inline; agent supply-chain governance for the external MCP servers and tools autonomous agents call; and granular financial and operational tracking with custom budgets, automated alerts, and daily anomaly digests.

Browser vs desktop, company: for browser-based AI use — an employee using a web-based assistant — Atlas can inventory that activity through existing telemetry, so shadow-AI use is seen. Granular inline control is applied at the desktop layer through Atlas AIDR.

Context the company cites, not a desk survey: 76% of employees now use AI at work, up from 30% two years ago, according to McKinsey as eSentire quotes it. The company says most organizations still cannot answer who is using AI, what data it is accessing, where that data is going, and what it costs, and that they lack audit-ready documentation.

Named voice on the release: James C. Foster, chief executive officer, eSentire. He talks about an accountability and security gap — enterprises racing to put AI to work without being able to say what their AI agents are doing with proprietary data, or what those agents cost. His quotes are color only and stay in Sources.

Partner quote, company-attributed: Nim Nadarajah, CISO and managing partner at CriticalMatrix, an eSentire partner. He says you can delegate work to AI but not accountability, and that clients need a record of what agents access plus the ability to enforce boundaries. The full quotes stay in Sources.

Platform context, still company: this acquisition builds on June 2026 Atlas Preempt — continuous, AI-led autonomous penetration testing and attack-path validation — and July Atlas Response, email security, and patch management. ESentire says it was named a Leader in the 2026 IDC MarketScape for Worldwide MDR Services for the midmarket, and that it protects more than 2,000 organizations across more than 80 countries and over two million endpoints. MDR is managed detection and response — outsourced security monitoring.

Availability, company: Atlas AIDR is generally available to eSentire customers and partners within the Atlas Platform. File GA as eSentire’s.

Plain English for the rest of the card: MDR = managed detection and response, meaning outsourced security monitoring. PII = personally identifiable information. MCP = Model Context Protocol-style tool servers that agents call. prompt injection = tricking an AI with crafted text to bypass its rules. SIEM = security information and event management, the log-and-alert system a security team already runs. shadow AI = staff using AI tools the security team has not approved.