Q3 2026 threat intelligence report

AegisAI Q3 2026 State of the AI Attack Report

Phishing rose 49% this quarter. Attackers are sending from compromised accounts and delivering through the calendar, and AI writes the lures and pushes one template to thousands of organizations in minutes.

Free report. Q3 2026.

Get the Q3 2026 report

The full findings, the data behind them, intrusion-set case studies and the methodology.

Thanks for submitting the form.

Free · Instant download

100M+

emails analyzed, across 1M+ protected mailboxes

38x

more AI Dragnet attacks sent from hijacked accounts

73%

of tracked organizations hit by one calendar worm

Built by the team behind Google Safe Browsing and reCAPTCHA. Backed by Battery Ventures, Accel and Foundation Capital.

Download our Q3 2026 report

The full findings, the data behind them and how we measured it, from more than 100 million emails.

Inside the report

What the Q3 2026 report covers

Compromised accounts: attackers send as the people you trust
Calendar attacks: phishing that lands with no click and outlives the email
AI Spearphishing and AI Dragnet attacks, with evasion, DMARC and brand breakdowns
+49%
rise in confirmed phishing, relative to overall email volume
38x
AI Dragnet attacks from hijacked accounts, 0.2% to 7.7%
73%
of tracked organizations reached by one calendar-invite worm
AEGIS IN NUMBERS

Built to run beside the mailbox you already have.

The 90% figure is measured by Aegis in customer environments, compared with rule-based filtering.

90%
Fewer false positives than rule-based filters
0
Changes to your email routing (MX records)
5 min
To connect Microsoft 365 or Google Workspace
FAQ

Questions about the report

How the data was collected, and what you get.

More than 100 million emails sent to over 1 million mailboxes protected by AegisAI, from July 1 to September 2, 2026. The AegisAI threat research team confirmed every phishing verdict in the report.

Phishing that blasts one AI-written template at as many organizations as possible, often within minutes. Its share of all confirmed phishing grew 11.6x this quarter, from 0.5% to 5.8%.

AegisAI's classifier scores each confirmed message on stylistic signals common in language-model output, plus hard artifacts such as unfilled template placeholders. The methodology section explains how it works and where its limits are.

The executive summary, the two key issues, breakdowns by attack type, intrusion-set case studies and the full methodology.

Fill in the form at the top of this page and the download opens right away. It is free.

Yes. Connect Microsoft 365 or Google Workspace by API, with no MX change, and Aegis runs in monitoring mode next to your current stack. Book a demo to get started.

See these attacks in your own tenant.

Connect a tenant, run in monitoring mode, and compare against what your current stack delivered.

Five-minute connect · monitoring mode · disconnect any time without touching mail routing