
score
Visiting through this link may earn us a commission if you sign up, at no extra cost to you. It doesn't change our review, our lowest editorial scores go to vendors who pay us the most for placement just as often as ones who pay us nothing.
Abnormal AI (formerly Abnormal Security) is an email security platform, and it's included here because payroll diversion fraud, someone impersonating an employee or an executive by email and asking HR or payroll to change direct deposit details, is genuinely one of the most common ways small and mid-size businesses actually lose money to 'payroll fraud' in practice. It's not exotic, it's a convincing email and a rushed approval.
Abnormal doesn't rely on the traditional approach of scanning for malicious links or attachments, because these attacks usually don't have any, they're just well-written social engineering. Instead it models normal communication behavior for your organization and flags anomalies: a request coming from a personal email address, a lookalike domain, unusual urgency or tone, a banking-detail change request that doesn't match how that person normally communicates.
This is a security tool, not an HR or finance tool, so it needs to be evaluated and bought by whoever owns your email security stack, and it doesn't touch payroll data directly. It stops the fraudulent request from landing convincingly in the first place, which is a different (and arguably more important) layer than anything that audits payroll after the fact.
Where Abnormal sits now is the premium end of this category. There's no free tier, pricing is per-mailbox and quoted, and onboarding is a paid, hands-on engagement (professional services commonly run five figures). What that buys is a genuinely hands-off, high-accuracy behavioral engine. If you want the same class of detection but need to start free and are willing to tune it yourself, Sublime is the SMB-accessible alternative, and it's our pick in this category for exactly that reason.
Pricing
Contact the vendor for current pricing — Abnormal is premium, quote-only email security with no free tier, priced per mailbox on employee count plus paid onboarding (professional services commonly $5k-$25k+). Reported per-mailbox rates work out to a few dollars a month, so a 200-mailbox company should budget low five figures a year before services.
Features
- Behavioral modeling of normal internal and vendor email communication
- Detection of payroll diversion and direct-deposit-change social engineering
- Lookalike domain and compromised-account detection
- Vendor email compromise (VEC) detection for fraudulent invoice requests
- No dependency on malicious links/attachments to trigger detection
- Integrates with Microsoft 365 and Google Workspace
- Targets the single most common real-world payroll fraud vector for SMBs: social engineering by email
- Behavioral detection catches attacks with no malicious payload, which slip past traditional email security
- Sits upstream of payroll, stopping the fraudulent request before it's ever acted on
- This is an email security purchase, not a payroll or HR tool, different budget owner and evaluation process
- No visibility into payroll data itself, it only stops the request from being convincing
- Enterprise pricing and sales process, no public numbers to compare against
Any business that processes direct-deposit change requests by email or Slack and wants to stop social-engineering-driven payroll diversion before it happens.
Teams looking for a tool that audits payroll records themselves, this doesn't touch that layer.
Screenshots
Supported countries
Cloud email security works anywhere Microsoft 365 or Google Workspace is used, not limited to the countries listed above.
Integrations
Abnormal AI alternatives
Tools worth comparing before you commit.