Google and Mandiant’s M-Trends 2026 report confirms that vishing is now the second most common initial breach vector worldwide and the number one vector in cloud environments. AI-driven attack platforms are now capable of automating the entire social engineering call flow. The only effective defense is preparing employees to recognize and resist these calls before attackers succeed.
Why Does It Matter That Vishing Leads the Google/Mandiant M-Trends 2026 Report?
M-Trends report is based on more than 500,000 hours of frontline incident response investigations conducted by Mandiant during 2025, combined with telemetry from the Google Threat Intelligence Group (GTIG). When the report highlights a trend, that trend is already part of real-world enterprise threat models. The 2026 edition does not open with ransomware, zero-day exploits, or nation-state APTs. It opens with vishing.
Top 2025 vectors
| Attack Vector | Share of 2025 Investigations | Change vs. 2024 |
| Exploits (CVEs) | 32% | Stable (6th consecutive year at #1) |
| Voice phishing (vishing) | 11% | ↑ Significant increase |
| Prior compromise | ~10% | ↑ Up from #5 in 2024 |
| Stolen credentials | 9% | ↓ Down from 16% in 2024 |
| Web compromise | 8% | Stable |
| Email phishing | 6% | ↓ Down from 14% in 2024 |
| Insider threat | 6% | ↑ Up from 5% in 2024 |
The simultaneous decline of email phishing and the rise of vishing are not accidental. Attackers follow the path of least resistance. Email security controls have improved. Human trust in voice communication remains much harder to challenge. In cloud environments specifically, vishing became the single most common initial compromise vector, accounting for 23% of cloud-related incidents. That placed it ahead of stolen credentials (16%), email phishing (15%), and exploits (6%).
The report also makes an important distinction between two attack categories. Email phishing is classified as a non-interactive technical lure, while vishing is categorized as interactive human engagement. The operational difference matters because interactive attacks are described as “significantly more resilient against automated technical controls” and require fundamentally different detection strategies. A firewall cannot block a persuasive phone conversation.
Key takeaways from the M-Trends 2026 report
- Vishing is now the #2 initial infection vector across all Mandiant incident response investigations conducted in 2025.
- Email phishing dropped to 6% of intrusions in 2025, compared to 14% in 2024, while vishing increased significantly.
- UNC3944 (linked to Scattered Spider) has been running vishing campaigns since at least early 2022, targeting help desks to bypass MFA protections.
- UNC6040 used voice phishing to trick victims into authorizing attacker-controlled SaaS applications, later used in ShinyHunters-branded extortion campaigns.
- The report emphasizes that interactive voice attacks are significantly more resistant to automated security controls than email phishing.
- The growing “hand-off” model allows one threat actor to gain access via vishing and rapidly sell that access to ransomware operators, sometimes in under 30 seconds.
What Do Real Vishing Attacks Look Like?
The M-Trends 2026 report references UNC3944 and UNC6240 as representative examples, but those campaigns are already widely documented. More recent reports demonstrate an even higher level of sophistication in vishing operations. AI-driven calling infrastructure is now commercially available, and nation-state actors are using fake video meetings to collect voice recordings for future impersonation attacks. Two recent cases are especially illustrative.
Scenario 1: The TOAD Attack Kit — AI Automates the Entire Call
Threat researchers at Abnormal AI published an April 2026 analysis of a cybercrime platform called ATHR. The platform is sold on underground forums for $4,000 plus 10% of profits. It is one of the clearest examples to date of how vishing has become industrialized.
ATHR combines the full attack chain into a single browser-based platform:
1. Lure delivery: A built-in mailer generates spoofed brand notification emails impersonating Google, Microsoft, Coinbase, and Binance. Attackers can customize fields such as account lock time, failed login attempts, last login location, and IP address. Each email is tailored to appear believable during casual inspection.
2. Callback routing: When a victim calls the number embedded in the email, ATHR routes the call either to a human operator or to an AI voice agent running on Asterisk WebRTC infrastructure, the same type of infrastructure used by legitimate call centers.
3. AI-driven social engineering: ATHR’s AI vishing agents follow a structured ten-step script. They verify the callback, describe fabricated suspicious activity, and extract six-digit verification codes without involving a live operator. One operator can run simultaneous campaigns against multiple brands.
4. Live credential harvesting: While the conversation is taking place, ATHR’s phishing panels capture credentials in real time. Operators can monitor each victim session live and redirect targets to specific phishing pages during the call itself.
What makes this especially important from a security perspective is that large-scale vishing campaigns no longer require highly skilled social engineers. The platform reduces operational cost and dramatically increases scalability. Help desk staff, finance teams, and IT administrators can now be targeted systematically.
Scenario 2: UNC1069 — Fake Meetings, Voice Harvesting, and North Korean Infrastructure
Researchers at Validin published a detailed April 2026 analysis of UNC1069, a North Korean threat actor associated with Bluenoroff. The group targets cryptocurrency companies, Web3 organizations, and financial services providers.
The attack chain
1. Attackers create fake venture capital personas and approach targets through LinkedIn and Telegram, often using previously compromised accounts to appear legitimate.
2. Meetings are scheduled through Calendly links leading to fake video conferencing platforms designed to mimic Zoom, Google Meet, and Microsoft Teams at the infrastructure level.
3. During the meeting, attackers claim the victim’s microphone or webcam is malfunctioning. They create urgency and present a ClickFix-style prompt instructing the target to paste and run a command to “resolve the issue.”
4. Critically, these fake meeting platforms record the victim’s audio and video. The recordings are later reused in additional campaigns to impersonate real individuals, including deepfake versions of executives.
This goes beyond traditional phishing. It is a long-term, multi-session social engineering operation where the voice channel serves both as the attack vector and as an intelligence-gathering mechanism. Mandiant tracks this activity cluster as part of the broader vishing and prior-compromise trends discussed in M-Trends 2026.
Why Are These Attacks Working?
Both scenarios rely on the same core advantage: exploiting trust during real-time human interaction.
Email phishing depends on convincing someone to click a link without inspecting it carefully. Vishing introduces a human voice into the interaction — responsive, contextual, and patient. Email gateways can block malicious attachments. They cannot detect a convincing pretext delivered slowly and naturally with a regional accent.
How Do You Defend Against Vishing? Train People to Recognize the Call
Technical controls can stop known malware. They cannot stop someone posing as an IT employee and convincing a CFO to authorize a wire transfer because “the CEO is unavailable in a meeting.” The only scalable defense is behavioral conditioning through realistic simulation. Employees need exposure to the same call dynamics attackers use, but within a controlled environment that includes immediate learning feedback.
How Simulated Vishing Attacks Work
A well-designed vishing simulation program typically serves two separate purposes.
1. Audit and risk assessment
A realistic vishing campaign is launched against the organization without advance warning or training prompts. Behavioral data is then collected:
- Which teams comply?
- Which roles reveal credentials under pressure?
- Which attack scenarios are most effective?
This establishes a measurable baseline and identifies which groups require targeted training.
2. Awareness training
Simulations are also used as a learning mechanism. When a user fails a simulated vishing call, the platform immediately delivers contextual feedback at the moment of failure rather than months later in a generic awareness presentation.
What Makes a Vishing Simulation Realistic Enough to Matter?
Three variables determine whether a simulation produces actionable insights or meaningless noise.
Pretext: The scenario must reflect real attacker tactics: IT help desk impersonation, bank fraud alerts, executive impersonation, or HR callbacks. Generic scenarios rarely produce meaningful risk assessment results.
Voice: The caller’s voice must align with the scenario. AI-powered simulations can use voice cloning to imitate known colleagues or executives — exactly the type of attack scenario that security leaders increasingly worry about and that groups like UNC1069 are already using.
Phone number: Spoofed phone numbers that appear to originate from the organization itself or from trusted vendors significantly increase compliance rates. Simulations using generic phone numbers test weaker scenarios than the ones attackers deploy in reality.
When Should AI-Powered Vishing Simulation Be Used?
Two scenarios stand out clearly.
| Use Case | Why AI Simulation Is Required |
| Large-scale campaigns (100+ users) | Human operators cannot scale efficiently. AI voice simulation enables organization-wide testing within a single campaign cycle. |
| Advanced threat scenarios | Testing resilience against voice cloning and deepfake impersonation requires the actual technology rather than scripted human callers. |
Arsen’s vishing simulation platform uses real-time voice AI and voice cloning to replicate the same attack techniques described in M-Trends 2026. Campaigns can target high-risk groups such as finance teams, IT staff, and executives, or run across the entire organization. Results are tracked at both the individual and team levels and integrated into ongoing security awareness workflows.
What Should a Vishing Simulation Program Include?
- A scenario library covering IT help desk impersonation, executive fraud, HR callbacks, vendor impersonation, and MFA bypass attempts
- Multiple voice options, including standard AI voices, regional accent matching, and executive voice cloning
- Spoofed caller IDs matching internal or trusted vendor numbers
- Real-time call tracking and compliance metrics
- Immediate post-call learning feedback for users who comply
- Reporting dashboards providing CISO-level visibility by team, role, and location
If you are interested in Arsen, a phishing, vishing, and smishing simulation platform, and would like to get a free trial version, please leave your contact details in the form below.
