Agenticum · Thoughts · 7 min read

AI Cybersecurity and Fraud Detection Tools: A 2026 Guide for UAE and Saudi Buyers

This guide maps AI security tools to the layer each one protects, from identity and email to source code. It also covers AI fraud products that card networks and specialist vendors offer to banks and insurers. Prices are listed in USD, with AED and SAR conversions. Performance figures are attributed to their sources.

Nikolay Zuykov · Agenticum · 29 October 2026

What counts as an AI cybersecurity tool in 2026

In this guide, an AI cybersecurity tool is a security product where machine learning or an AI assistant does part of the detection or investigation work. Some vendors describe their products as agentic. CrowdStrike calls Charlotte AI "an agentic AI security analyst" built into its Falcon platform. SentinelOne describes Purple AI, which is embedded in its Singularity Platform, in almost the same words.

Microsoft cites a figure of 78 trillion security signals analysed per day, taken from its Digital Defense Report 2024. It also says its identity risk detections are built on analysing trillions of signals each day.

The FBI's Internet Crime Complaint Center (IC3) received 1,008,597 complaints in 2025. Reported losses were USD 20.877 billion, 26% more than in 2024. More than 22,000 of those complaints contained AI-related information, with adjusted losses above USD 893 million. Businesses also reported losing more than USD 30 million to business email compromise (BEC) scams that involved AI. Investment fraud was the largest loss category, followed by BEC and tech support scams.

AI cybersecurity tools by category: comparison table

The table shows list prices from vendor pages. AED and SAR amounts use the fixed rates explained further down. "Not cited here" means this guide does not cite a price or licensing detail for that product.

Category	Example tool	What it does	Licensing	Published price (USD / AED / SAR)	Source
AI security analyst	CrowdStrike Falcon with Charlotte AI	CrowdStrike describes Charlotte AI as an agentic AI security analyst built into the Falcon platform. This guide does not specify which plans include Charlotte AI.	Per device, billed annually	Falcon Go 59.99 / 220.31 / 224.96; Falcon Pro 99.99 / 367.21 / 374.96; Falcon Enterprise 184.99 / 679.38 / 693.71; Falcon Complete: quote only	CrowdStrike pricing
AI security analyst	SentinelOne Purple AI	Guides investigations, explains risk and recommends next actions	Not cited here	Not cited here	SentinelOne
SIEM/SOC	Microsoft Sentinel	Ingests alerts from Defender XDR and the other Defender products free of charge	Pay-as-you-go per GB, or commitment tiers from 100 GB/day	Per-GB rate not cited here; 31-day free trial on the first 10 GB/day	Microsoft Learn
SOC assistant	Microsoft Security Copilot	Billed in Security Compute Units (SCUs)	Provisioned SCUs plus usage-based overage	4.00 / 14.69 / 15.00 per provisioned SCU per hour (Microsoft estimate); 6.00 / 22.04 / 22.50 per overage SCU	Microsoft pricing
Email	Defender for Office 365 P1 / P2	P1: protection against zero-day malware, phishing and BEC. P2 adds phishing simulation, hunting, and Automated Investigation and Response	P1 in suites such as E3 and Business Premium; P2 in suites such as E5	Standalone price not cited here	Microsoft Learn
Identity	Microsoft Entra ID Protection	Detects identity risks such as password spray and leaked credentials	Requires Entra ID P2 licences	Not cited here	Microsoft Learn
Cloud	Wiz	Agentless API connection; the Security Graph analyses relationships between cloud technologies to prioritise risk	Not cited here	Not cited here	Wiz
AppSec	GitHub Code Security with Copilot Autofix	Generates fix suggestions for CodeQL code-scanning alerts	Per active committer per month; Autofix included on Team and Enterprise	Code Security 30 / 110.18 / 112.50; Secret Protection 19 / 69.78 / 71.25	GitHub plans

AI security operations: what the SOC assistants really do and where they stop

SentinelOne says Purple AI reasons across security data to guide investigations, explain risk and recommend next actions within SOC workflows.

Check what each accuracy claim is measured against. CrowdStrike says Charlotte AI Detection Triage was benchmarked at over 98% accuracy. The comparison is with decisions made by CrowdStrike's own Falcon Complete MDR team. That makes it a vendor benchmark, not an independent test.

GitHub states the limits of its own tool. Copilot Autofix suggests fixes for CodeQL alerts, and GitHub warns that a suggestion may fail to fix the vulnerability or may introduce new vulnerabilities. Review each suggested fix the same way you would review any pull request.

Security Copilot is billed in Security Compute Units (SCUs). In its billing example, which Microsoft labels an estimate, Security Copilot costs USD 4 per provisioned SCU per hour. Overage SCUs cost USD 6 each and are billed on usage. Eligible Microsoft 365 E5 and E7 customers get 400 SCUs per month for every 1,000 user licences, up to 10,000 SCUs per month. Microsoft says activation is being rolled out in phases.

AI fraud detection in banking and card payments

Visa says its AI fraud models give each transaction a risk score, typically from 0 (low risk) to 99 (high risk). Visa completed its acquisition of Featurespace, which develops real-time AI payments protection technology, on 19 December 2024. Mastercard says its Decision Intelligence helps banks score and approve 143 billion transactions a year. Feedzai positions itself as an AI-native fraud and financial crime prevention platform for banks and fintechs.

Area	Provider and product	Published figure	Status
Cards	Visa Decision Manager	98.7% of transactions resolved automatically by AI; an estimated USD 33 billion in potential fraud losses prevented in 2023	Vendor-reported
Cards	Mastercard Decision Intelligence	Generative AI enhancements raise fraud detection by 20% on average and by up to 300% in some cases	Vendor modelling (initial, 2024)
Banking	Feedzai	62% more fraud detected than a tier-1 bank's previous solution	Vendor-reported
Insurance	Shift Technology	4B+ policies, claims and documents analysed; insurers in 30+ countries	Vendor-reported

AI insurance fraud detection

Shift Technology says it has analysed more than 4 billion policies, claims and documents. Shift says it serves property and casualty, travel, health, workers' compensation and life insurers in more than 30 countries. When you compare a claims tool with a payment tool, ask which claim and document data the model needs from you and where that data will be processed.

UAE and Saudi rules that shape AI fraud prevention

Saudi Arabia: SAMA

Counter-Fraud Framework. SAMA expects member organisations to operate at maturity level 3 or higher. At level 3, "fraud detection system capability is implemented and embedded". The framework also expects counter-fraud data analytics that can analyse large volumes of transactions.
Counter-Fraud Fundamental Requirements. These apply to finance companies and payment service providers. They replace the Anti-Fraud Rules for Finance Companies and took effect on 13 April 2026.

Saudi Arabia: NCA

The National Cybersecurity Authority updated its Essential Cybersecurity Controls as ECC 2-2024 to protect the information and technology assets of national entities. The English document lists 4 main domains, 28 subdomains, 108 main controls and 92 subcontrols.

UAE

Gulf News, citing Central Bank guidelines, reported that UAE banks began phasing out SMS and email OTPs on 25 July 2025 in favour of in-app authentication. It also reported that SMS OTPs would be discontinued entirely by March 2026. This is press reporting; no first-party circular is cited here.

In practice, a SAMA member organisation should be ready to show that its fraud detection capability is implemented and embedded, and that its counter-fraud analytics can analyse large volumes of transactions. A UAE business that used SMS codes to verify customers should confirm that its identity and fraud tools work with in-app authentication.

Prices in AED and SAR, and how to choose

AED figures use USD 1 = AED 3.6725, the rate on the Central Bank of the UAE's official rate sheet. SAR figures use USD 1 = SAR 3.75. These conversions use fixed exchange rates. Ask each vendor which currency it invoices in.

Identify the security capabilities you need, then compare the relevant products in the table. Before buying two products in the same category, check whether their coverage overlaps.
Check what you already own. Microsoft 365 E5 includes Defender for Office 365 P2 and, for eligible customers, a monthly SCU allowance. E3 and Business Premium include P1. Defender XDR alerts flow into Sentinel free of charge, but raw logs for some data types are paid.
Check what each metric is measured against. CrowdStrike reports over 98% accuracy for Charlotte AI Detection Triage against decisions made by its own Falcon Complete MDR team. Feedzai's 62% is measured against one bank's previous solution. Mastercard's figures come from initial modelling.
Require human review of suggested fixes and ask how automated decisions are recorded. GitHub itself says Autofix suggestions must be reviewed. Ask every vendor how its automated decisions are logged, so that you can explain them to an auditor or regulator.
Run a trial before committing. Sentinel's 31-day trial covers the first 10 GB/day of Analytics ingestion. Commitment tiers start at 100 GB/day, so measure your real daily volume during the trial before you choose between pay-as-you-go and a tier.
Plan for platform changes. After 31 March 2027, Sentinel will be available only in the Microsoft Defender portal, not the Azure portal.
Define how employees may use AI with company data. Our guide to a company AI policy covers employee use of AI with company data.