Abnormal AI Governance
by Abnormal AI
Shadow AI discovery, AI agent inventory and AI spend tracking built from email, identity and OAuth signals
Abnormal AI Governance is a shadow AI discovery and governance product for security and IT teams that need to know which AI tools, agents and subscriptions their employees actually use. It reads email, identity and OAuth signals from Microsoft 365 or Google Workspace in read-only mode, so a team can inventory AI use without deploying an agent or proxy.
Abnormal AI Governance is the AI security module of Abnormal AI's Behavioral Security Platform, the email and identity security company founded in 2018. Abnormal unveiled it as part of the platform in July 2026, marked it generally available around Black Hat USA in August 2026, and on November 3, 2026 announced that it can be bought on its own, without Abnormal's Inbound Email Security product (the release dates standalone purchase to October 30, 2026). The product connects to Microsoft 365 through the Microsoft Graph API, or to Google Workspace, in read-only mode, and infers AI use from signals that already pass through those systems: signup and invoice emails, OAuth grants and identity data. From that it builds four views. AI tool discovery classifies each tool as sanctioned, tolerated or unsanctioned and records who uses it, what it can access and when it first appeared. The AI agent inventory finds agents on connected platforms such as Microsoft Copilot Studio and flags dormant, orphaned or overprivileged ones. AI chat insights scan what employees share with AI assistants, including uploaded files and images, for credentials, personal data and financial data, ranked by severity. Spend and compliance reporting consolidates AI subscriptions and invoices, including spend outside procurement, and maps findings to compliance frameworks for audit and board reporting. Abnormal reports that in its trials two of every three AI tools in use had never been approved by IT or security; that figure is vendor trial data. Licensing is per user, counted by mailbox. Abnormal says it protects more than 5,000 organizations, including 30% of the Fortune 500, as of July 31, 2026.
A CISO or security team at a Microsoft 365 or Google Workspace shop that needs a shadow AI and AI agent inventory before it can write or enforce an AI use policy.
A read-only inventory of AI tools, agents, sensitive chat data and AI spend across the company, built from signals the mail and identity systems already hold.
At a Glance
- Category
- Governance & Security
- Pricing
- Subscription, Contact for pricing
- Target Market
- CISOs, CIOs, Security Operations Teams, IT Governance and Compliance Teams
- Deployment
- Cloud-only, API-based
- Founded
- 2018
- Customers
- More than 5,000 organizations on the Abnormal platform, including 30% of the Fortune 500 (company figure, as of July 31, 2026)
Key Features
- ✓AI tool discovery
Finds AI tools from signup emails, OAuth grants and identity signals, then classifies each as sanctioned, tolerated or unsanctioned with its users and access.
- ✓AI agent inventory
Catalogs agents across connected platforms such as Copilot Studio and flags dormant, orphaned or overpermissioned agents that still hold access.
- ✓AI chat insights
Scans data shared with AI assistants, including uploaded files and images, for credentials, personal and financial data, and ranks findings by severity.
- ✓AI spend tracking
Consolidates AI subscriptions and invoices, breaks token and subscription cost down by tool and team, and compares month-end forecasts with budgets.
- ✓Compliance mapping and audit trail
Maps findings to compliance frameworks with controls and evidence, and records every approval decision for audit and board reporting.
- ✓Owner notification workflow
Can automatically ask the person responsible for a flagged tool or grant to justify it, and route the answer to security for review.
Capabilities
Use Cases
- •Shadow AI baseline before an AI policy
A security team connects Microsoft 365 read-only and gets a list of every AI tool in use, with owners, before drafting an acceptable use policy.
- •Agent sprawl cleanup
An enterprise finds hundreds of dormant Copilot Studio agents still holding key permissions and retires them, shrinking the attack surface agents create.
- •Sensitive data exposure review
Compliance reviews severity-ranked findings of credentials and personal data pasted into AI chats and targets training at the teams responsible.
- •AI spend consolidation
Finance and IT surface AI subscriptions bought on expense outside procurement and consolidate them onto sanctioned enterprise contracts.
Ideal For
Best For
- ✓Building a first inventory of sanctioned and unsanctioned AI tools without deploying endpoint agents or a proxy
- ✓Finding dormant, orphaned or overprivileged Copilot Studio and other platform agents that still hold permissions
- ✓Detecting credentials, personal data and financial data shared with AI chat assistants, including in uploaded files
- ✓Consolidating AI subscription and invoice spend, including purchases made outside procurement
- ✓Existing Abnormal email security customers who can switch the module on from the Abnormal AI App Store
Not Ideal For
- ✗Organizations not on Microsoft 365 or Google Workspace, since discovery is built from those email, identity and OAuth signals
- ✗Teams that need inline, real-time blocking of prompts or agent actions at runtime; the product describes discovery, classification, owner notification and reporting rather than in-path enforcement
- ✗Buyers who want published list pricing; licensing is per mailbox and quoted through sales
Deployment
Market & Ratings
More than 5,000 organizations on the Abnormal platform, including 30% of the Fortune 500 (company figure, as of July 31, 2026)
Market Analysis
Pros
- ✓Read-only connection to Microsoft 365 or Google Workspace, with no new infrastructure to deploy
- ✓Finds AI spend from invoice emails, which network-based shadow AI tools do not see
- ✓Backed by a large existing customer base and a security program with SOC 2 Type II, ISO 27001, ISO 42001 and FedRAMP Moderate
Cons
- ✗Discovery depends on email, identity and OAuth footprints, so AI use that leaves none of those signals may go unseen
- ✗No independent reviews of the AI Governance module yet; G2, Hacker News and trade-press coverage found during research were about the email product or the company's funding
- ✗PeerSpot reviewers of the wider Abnormal platform cite limited third-party integrations and asynchronous API issues
- ✗Vendor statistics on unapproved tools come from Abnormal's own trials, and its customer counts differ between its own pages (4,500+ versus 5,000+)
Pricing
AI Governance (standalone)
Contact for pricing
- ✓Licensed per user, counted by mailbox
- ✓AI tool discovery, agent inventory, chat insights, spend and compliance reporting
- ✓Free evaluation of your own environment on request
No list price is published. AI Governance is licensed per user, with mailboxes as the proxy for user count, and is quoted through sales. Abnormal offers a free evaluation that it says returns findings from your own environment within about a week. Existing Abnormal customers activate it from the Abnormal AI App Store. PeerSpot reviewers of Abnormal's email product describe per-user pricing as competitive, with one citing just over $60,000.
Security & Compliance
Sources
This page was written from 7 sources, 3 on domains other than abnormal.ai.
- 1.abnormal.ai — abnormal ai makes ai governance available as a standalone prvendor
- 2.abnormal.ai — standalone ai governancevendor
- 3.abnormal.ai — ai governancevendor
- 4.abnormal.ai — trustvendor
- 5.securityweek.com — abnormal security raises 250 million at 5 1 billion valuatio
- 6.peerspot.com — abnormal security reviews
- 7.hn.algolia.com — search
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Akeyless Agentic Runtime Authority
Intent-based runtime access control for AI agents: credentialless access, per-action policy checks and a session kill switch
Darktrace / SECURE AI
Behavioral monitoring of enterprise AI use: shadow AI blocking, prompt risk analysis and AI agent identity mapping
Pillar Security
AI security platform for discovering, red-teaming and guarding AI agents, MCP servers and coding agents, being acquired by Vanta
Bay
AI agent security platform that discovers every agent and MCP server and blocks risky agent actions at runtime