Test your cyber skills

Insights/ Threat Landscape

Phishing left the inbox

One in three phishing attacks now arrives outside email. The evidence behind the shift, and the six channels your awareness program needs to cover in 2026.

Mono Training team14 min read
Executive summary

Phishing is no longer an email-only problem. One security vendor now counts one in three phishing attacks arriving through channels other than email, and incident responders see voice calls, text messages, LinkedIn, WhatsApp, QR codes and Microsoft Teams in their cases. This article examines the evidence behind the shift, what the Australian data shows, the six channels an awareness program needs to cover in 2026, and why most automated training platforms structurally can't follow.

Data sourcesASD's ACSC Annual Cyber Threat Report 2024 to 25 · CrowdStrike 2025 Global Threat Report and Threat Hunting Report · Cisco Talos Q1 2025 Incident Response Trends · Push Security, 2025 Top Phishing Trends · Joint CISA/FBI/ASD/AFP/NCSC advisory on Scattered Spider (July 2025)Figures reflect publicly reported research as of October 2026. Findings evolve. Verify current studies before citing externally.

01The evidence, not ambiguous

Phishing has left the inbox, and the data agrees

The headline statistic here isn't a rounding error. It sits inside a converging body of evidence from security research firms, government cyber agencies, and incident responders, all making the same point from different angles. Email remains the largest single channel, but it's now one of several, and the others are growing far faster than email is.

One in three phishing attacks now arrives through channels other than email.

That figure comes from Push Security, a browser security vendor, in its late 2025 research on phishing trends, and covers channels such as social media, search engine results and messaging apps. Vendor data deserves caution, but it doesn't stand alone: a set of corroborating numbers, each from a different vantage point, describe the same structural change.

442%

increase in voice phishing between H1 and H2 2024. By mid 2025, CrowdStrike had already seen more vishing than in all of 2024.

CrowdStrike, 2025 Global Threat Report
60%

of phishing-related incident response engagements in Q1 2025 (more than 60%) involved vishing, the most common type of phishing seen.

Cisco Talos, Q1 2025
82%

of phishing sites analysed now target mobile devices.

Zimperium, 2024

02The national picture

The Australian data: ASD's ACSC

ASD's ACSC confirms the same pattern in its Annual Cyber Threat Report 2024 to 25, published October 2025. Phishing, which the report treats as a type of social engineering, was recorded as an initial access technique in 38% of the incidents ASD's ACSC responded to in FY2024 to 25, and it was the top technique in both government and industry incidents.

More operationally, the report notes that information-stealer malware is commonly distributed through SMS or email phishing, and its advice for a suspected social engineering attempt starts with the phone: do not engage, hang up.

The one-in-three statistic is the global view. The ACSC data is what a CISO or risk committee member will recognise as the national reality. Both point to the same place.

03One category, six decisions

Six channels, six different dynamics

"Phishing" as an undifferentiated term is losing its usefulness. A program that covers "phishing" without distinguishing between channels is training people on a generic category, not on the specific decision they'll actually face. Each channel below has its own trust dynamic and its own teachable moment.

01 · Smishing

SMS-based phishing

Info stealer delivery route (ASD)

Fake delivery notices, bank alerts, ATO impersonation, fake MFA prompts. SMS feels more personal and more trusted than email, which is exactly why it works.

02 · Vishing

Voice-based phishing

Up 442% H1→H2 2024

The canonical pattern is help-desk impersonation: calling the help desk pretending to be an employee, or calling an employee pretending to be IT.

03 · Messaging apps

WhatsApp, Signal, Telegram

FBI and CISA warning, March 2026

Russian-linked campaigns posing as app support to hijack Signal and WhatsApp accounts, flagged by the FBI and CISA. The trust dynamics that make these apps useful for work make them useful for attackers.

04 · Social DMs

LinkedIn as the executive channel

Leading non-email channel (Push Security)

A compromised peer account sends an investment, recruitment, or partnership approach. Because it's from a trusted network, it bypasses the mental model used to screen cold contact.

05 · Quishing

QR code phishing

Hidden in PDFs and posters

Increasingly embedded in PDF attachments to evade URL scanning, and the page that loads sits outside every enterprise control the organisation has invested in.

06 · Collab platforms

Microsoft Teams and Slack

Help desk impersonation on Teams

Attacks typically come from a compromised internal account or federation exploit. The message lands in an interface people treat as internal. Scepticism drops accordingly.

04Critical infrastructure, live

A major Australian breach started at a call centre, not an inbox

In July 2025 Qantas, Australia's flagship airline, disclosed that an attacker had targeted one of its call centres and got into a third-party customer service platform holding records for about six million customers; it later confirmed 5.7 million were affected. Security researchers linked the attack to the voice phishing campaign against users of Salesforce-based platforms, and to the groups tracked as Scattered Spider and ShinyHunters. Qantas has not named an attacker.

The reported entry point was a call centre and the platform behind it, the kind of target this group of attackers reaches by phone.

The group's methods are documented in the joint advisory issued in July 2025 by CISA, the FBI, ASD's ACSC, the AFP, and the UK's NCSC. The advisory names help desk impersonation by phone, push bombing, SIM-swap attacks and MFA fatigue as ongoing techniques. Aviation is Australian critical infrastructure, and in this case the door was a contact centre, not a mailbox.

05Why now, specifically

Five forces, operating together

None of this is accidental. The move from an email-dominated threat surface to a multi-channel one has specific, reinforcing causes, and they're still accelerating.

01

Email security actually worked

Two decades of gateway scanning, DMARC, DKIM, and anomaly detection made email the most defended surface. Attackers moved to surfaces without those defences.

02

Other channels aren't screened

Social DMs, WhatsApp, SMS, and Teams chat travel over infrastructure that doesn't inspect content. A LinkedIn DM from a real account looks legitimate by design.

03

AI collapsed personalisation cost

Scrape a LinkedIn profile, clone a voice from three seconds of audio, generate a personalised approach at near-zero cost. AI-generated spear phishing reportedly achieves far higher click rates than human-written lures.

04

Mobile is least monitored

Smaller screens hide URL previews. Most organisations have far less security telemetry on the employee phone than on the employee laptop.

The fifth force: collaboration tools were never designed with anti-phishing as a primary control. Threat actors are increasingly using Teams and Slack for initial access and delivery, inside environments people treat as inherently trusted.

·Key takeaways

What to take into your next risk conversation

01Email is no longer the whole picture

Voice phishing is up 442%, and in Cisco Talos's Q1 2025 cases vishing was the most common form of phishing. A programme built for email alone leaves a growing share of attacks untrained for.

02Australian data supports the same conclusion

Phishing was the top initial access technique in ASD's incidents, and info stealers commonly arrive by SMS as well as email.

03Critical infrastructure is the live case study

A major Australian breach in 2025 started at a call centre, not an inbox. The July 2025 joint government advisory documents the playbook.

04Architecture determines what a program can teach

Email-gateway platforms were built for one channel. Human-led, intelligence-driven programs can cover six, because the channel isn't the constraint.