Should You Still Be Afraid of Vishing? The 2026 Evidence — and 10 Simulation Solutions to Test Your Defenses

Key Takeaways

• AI has made vishing scalable. Targeted campaigns that previously reached roughly 50 firms can now reach 1,000, driven by automated persona research and synthesized voices built from short audio samples.
• Vishing is now the second-most common initial infection vector, accounting for 11% of cases in Mandiant’s M-Trends 2026 report, ahead of email phishing at 6%.
• Caller ID, most MFA implementations, and annual awareness training each have a documented gap against AI-enabled voice attacks, including help-desk social engineering calls.
• The August 2026 hedge-fund wave confirms the threat is active, coordinated, and targeting organizations with mature security teams.
• Vishing simulation builds behavioral reflexes around verification, refusal, escalation, and reporting that technical controls and policy documents cannot. The comparison covers 10 platforms with that capability.

The August 2026 Hedge-Fund Vishing Wave

On August 5, 2026, Point72 Asset Management informed its investors that it had been attacked. The hedge fund’s initial review found no client information stolen, but the incident was still under investigation. Around the same time, according to reporting from Bloomberg, attackers made attempts against Two Sigma, Citadel, Millennium, and several private equity firms as part of the same coordinated campaign.

Two Sigma, which manages around $75 billion in assets, was direct about what happened: “Our security team responded quickly to an attempted vishing campaign targeting Two Sigma and other investment managers, and we have no indication of any impact to our data or our systems.” Spokespeople for Point72, Citadel, and Millennium declined to comment. No confirmed data breach has been tied to any of the named funds.

The voices on those calls were not human. They were synthesized to replicate the sound, cadence, and phrasing of people the targets recognized, then used to try to talk employees into revealing credentials or approving access they would not otherwise grant.

The scope matters as much as the technique. A structured campaign worked through some of the world’s most recognizable money managers. The pattern points to a deliberate operation rather than opportunism and fits what AI has made possible: one campaign reaching many high-value targets with well-researched calls convincing enough to get through a front-line employee.

The Financial Industry Regulatory Authority (FINRA) took notice. According to reporting from InvestmentNews, FINRA reached out to member firms about the attempted breaches. That response reflects an institution that has been watching this threat build. FINRA launched its Financial Intelligence Fusion Centre in March 2026 specifically to give financial firms a channel to share intelligence about fraud threats and coordinate responses. The hedge-fund wave was precisely the kind of event that facility was built for.

For a closer look at the attack mechanics and the specific defenses that held, see our full breakdown of the August 2026 hedge-fund vishing wave.

Why AI Turned Vishing Into a Mass Weapon

Voice phishing has existed for years. What changed is the cost of doing it well.

Before AI-enabled voice synthesis, a convincing impersonation call required real skill. The attacker needed to sound the part, know the right details, handle objections smoothly, and maintain the pretense under pressure. That took either talented operators or extensive preparation, which naturally limited how many targets any campaign could reach.

Vinod Paul, president of Align Managed Services, which specializes in cybersecurity for hedge funds, described the shift plainly in his Bloomberg comments: “Before, they could attack 50 entities in a targeted attack — now they can do 1,000.” When an attacker can synthesize a convincing voice from a short audio sample (pulled from an earnings call, a conference talk, a podcast, or an intercepted phone call) and automate the research needed to build a plausible pretext, the marginal cost of the next target approaches zero.

Will Wilson, CEO of Antithesis, which helps companies find and fix vulnerabilities and is backed by quantitative trading firm Jane Street, put the broader implication directly: “The terrifying thing about modern-day AI systems is that they have commoditised this and made it possible to execute attacks at scale. Everybody will have to seriously level up. Otherwise, they are going to be in big trouble.”

AI didn’t just make vishing cheaper to scale; it broke two trust signals that employees have relied on for years. Caller ID has always been spoofable, so it was never a reliable signal. But a familiar voice used to be a reasonable proxy for identity. When a call sounds exactly like the CFO, most people’s instinct is to treat it as the CFO. That instinct is now a liability. When voice can be synthesized from publicly available audio, the phone becomes an unauthenticated channel, and any process that treats “they sounded like our colleague” as sufficient verification is exposed.

Vishing Moved Across Every Sector in H1 2026

The hedge-fund wave did not happen in isolation. Voice phishing has been the leading-edge social-engineering method across multiple industries throughout 2026, and the data confirms it has moved from a niche attack into a primary initial access path.

Mandiant’s M-Trends 2026 report, based on hundreds of thousands of hours of frontline incident investigations, found that voice phishing accounted for 11% of initial infection vectors in 2025, making it the second-most common, behind exploits at 32%, and ahead of email phishing, which fell to 6%. Most security programs were built around email even though voice has now overtaken it as the more frequently observed social-engineering path into organizations.

The incidents that support that finding span multiple sectors:

Law firms and professional services (January–May 2026). Google’s threat intelligence unit documented a wave of vishing attacks against U.S. law firms and professional services companies, attributed to a cluster designated UNC3753. The attack chain was direct: a vishing call impersonating IT support, a screen-sharing session, a remote-management tool installed, and data exfiltrated in under 30 minutes, all within a single business day. Some incidents went further, with individuals physically entering corporate offices posing as IT workers.

Financial SaaS (April 2026). In April, attackers compromised ADT’s Salesforce instance after using a vishing call to obtain Okta SSO credentials. With those credentials, they accessed and exfiltrated customer and prospect data. ShinyHunters claimed responsibility and reported over 10 million records. The access path was entirely social: no vulnerability exploited, just a phone call that convinced an employee to hand over the right credential.

UK retail (2026). Scattered Spider’s attacks on Marks & Spencer, Co-op, and Harrods followed a consistent pattern: call the service desk, impersonate an employee, claim a lost device, and request MFA removal and re-enrollment. At M&S, the downstream impact included DragonForce ransomware deployed across VMware ESXi systems, disruption to online orders and warehouse operations, and a market-cap impact estimated above £500 million.

SaaS platforms (early 2026). Google’s threat intelligence group documented the BlackFile cluster, designated UNC6671, targeting organizations across North America, Australia, and the UK. The initial access method was a vishing call to a personal cell phone, leading to a lookalike SSO portal and a real-time adversary-in-the-middle relay that captured credentials and registered an attacker-controlled MFA device. From there, the cluster moved through Microsoft 365 and Okta-connected SaaS, exfiltrating files at scale.

The New York State Department of Financial Services issued an industry letter in February 2026 warning of an ongoing cyberthreat campaign involving targeted vishing attacks. The letter was directed at financial institutions but described techniques that have since appeared across every sector above.

Attackers adapt the same techniques to whichever target list offers the best return: finance in August, with law firms and retail targeted earlier in the year. The technique travels; only the sector changes.

What Your Current Defenses Don’t Cover

Most security teams can point to an existing control for every part of a vishing attack. Having a control does not mean it covers this attack path.

Caller ID has always been spoofable, and AI didn’t change that. What AI did was lower the cost of making a convincing call to go with the spoofed number. Caller ID tells you what the attacker wants you to see. Any process that accepts a number displayed on a phone as identity verification has a gap.

Multi-factor authentication. Phishing-resistant MFA (FIDO2 security keys, passkeys) is structurally sound against credential phishing because there is no code to read aloud and no push notification to approve under pressure. That protection covers an attacker trying to steal a credential and replay it elsewhere, but not a help-desk social engineering call. The attack path used by Scattered Spider, UNC3753, BlackFile, and ShinyHunters calls the service desk, impersonates an enrolled employee, claims a lost or broken device, and asks the agent to remove the existing MFA factor and enroll a new one. The agent complies, and the attacker inherits the account. The MFA system worked exactly as designed; the process that manages it was the failure point.

SMS OTP and push notification-based MFA have two additional gaps against vishing: an attacker on a live call can instruct the target to read back the one-time code as it arrives, or pressure them to approve a push notification that just appeared. Neither of those paths requires stealing a credential. They only require keeping the target on the line long enough.

Annual awareness training. The Fortinet 2025 Security Awareness and Training survey found that 94% of organizations run regular security training, but only 6% achieve full completion, and 69% still report that employees lack adequate awareness. Knowledge of what vishing is (its definition, its red flags, its general anatomy) does not produce a refusal reflex when a confident, adaptive AI caller is applying real-time pressure. A Harvard-led study of 4,100 participants testing AI vishing scenarios found that 16.5% of subjects self-reported they would comply. The study also found that persuasion tactic design predicted compliance more reliably than whether subjects perceived the voice as human. Scenario design matters as much as audio quality.

Deploy phishing-resistant MFA on privileged accounts and executive-adjacent roles, and keep awareness training for baseline knowledge. Neither control trains the behavioral reflex needed when someone is on the phone under pressure: hang up, call back through a verified internal number, and escalate rather than act alone. Simulation addresses that behavioral layer.

What a Vishing Simulation Actually Tests

Security awareness training and vishing simulation both aim to reduce susceptibility, but they operate at different points in the process. Training delivers information about what vishing looks like and what employees are supposed to do. Simulation puts an employee on a call with a caller applying the same social-engineering tactics a real attacker would use: authority, urgency, familiarity, commitment escalation. What it measures is what the employee actually does under those conditions, not what they know when sitting at a desk.

The behaviors that matter are specific:

• Does the employee ask for a callback number, then verify it against an internal directory before calling back?
• Do they refuse to take sensitive action on an inbound call and escalate to a supervisor instead?
• Do they report the call promptly, even if they complied, so the security team can triage it?
• Do help-desk agents require identity verification before processing a credential or MFA reset?

A simulation that triggers on employees who fail those decision points, and immediately delivers targeted remediation, achieves something annual training cannot: it creates a learning moment at the point of failure, when the lesson is most likely to stick. The Harvard finding reinforces this. Because persuasion design predicts compliance more than voice realism, a simulation built around well-designed social-engineering scenarios will expose behavioral gaps even when employees know they’re being tested.

If you’re building the business case for adding vishing simulation to your awareness program, we’ve covered why CISOs are adding vishing simulation to their programs in a separate analysis.

What to Look for in a Vishing Simulation Platform

Not all platforms simulate vishing at the same level of realism or depth. Before evaluating specific vendors, align on what your program actually needs.

Conversation type. Vishing simulation falls into three categories: prerecorded or voicemail-based (fixed audio, no interaction), interactive template-based (scripted branching flows), and live adaptive AI (generative, real-time conversation that responds to what the employee says). Voicemail simulations build basic recognition. Live adaptive AI is required to rehearse the real-time pressure and objection handling that characterizes a skilled attacker.

Scenario control. Can you set the attack goal, caller persona, pretext, and urgency level yourself? Or are you limited to vendor-provided templates? Custom control lets you simulate the specific scenarios most relevant to your roles: a service-desk reset request, a finance wire approval, a credential capture dressed as an IT security check.

Voice options. Assess the preset voice library for language coverage relevant to your workforce, and check whether the platform supports custom voice cloning for executive impersonation scenarios. If it does, confirm what the consent and legal review requirements are: voice cloning carries jurisdiction-specific obligations.

Multi-channel chaining. Some of the most effective real-world attacks combine a call with a follow-up phishing email, or use an email as a pretext before calling. A platform that can coordinate both channels in one campaign reflects how attackers actually operate.

Vishing-specific metrics. Answer rate, call duration, and failure rate by scenario type give you a clearer picture of program performance than click-rate analogies. Tracking these over time reveals whether refusal behavior is actually improving.

Remediation. Automatic follow-up training triggered on simulation failure is the most effective intervention point. Confirm that training is framed around behavior change rather than punishment, and that the content is specific to the failure type.

Program safety. Cooling periods prevent the same employee from being targeted by the same scenario in quick succession, which would degrade data quality and erode trust. Confirm the platform has one and that it’s configurable.

10 Recommended Vishing Attack Simulation Platforms

The platforms below were selected based on documented vishing simulation capability, scenario design control, and relevance to security-awareness programs evaluating phone-channel coverage. All capability details come from vendor sources and may vary by plan, region, or release date. Validate them in a proof of concept before committing. For a deeper evaluation rubric and proof-of-concept checklist, see our vishing simulation software buyer’s guide.

Brightside

Brightside is an AI-era security awareness platform built around simulation realism across phishing, vishing, and deepfake scenarios. Its AI vishing simulator is one of the most fully featured self-serve tools in the market, designed for security teams that want direct control over how a campaign is built and what it tests.

The template creation workflow runs in five steps: attack goal, context, tactics, voice, and review. Admins select between a Voice Attack (call only) or a Hybrid Attack that coordinates a call with a trackable phishing email in a single campaign. The attack goal drives an AI-generated caller persona and opening message, with an optional auto-fill that populates caller name, role, and organization based on the objective. An AI-recommended strategy layer then suggests a combination of social-engineering tactics drawn from Pretexting, Curiosity Hooks, Fear/Threat, Commitment, Authority Impersonation, Social Proof, and Reciprocity, along with urgency level, conversational tone, and a plain-English explanation of the psychological reasoning behind the combination. Admins can accept the recommended strategy or build their own.

The voice library includes preset options across English, French, German, Italian, Polish, Spanish, Dutch, and additional languages provisioned on request. Custom voice cloning creates an executive-specific voice replica from a 1–2 minute recording. Before any campaign goes live, admins can test the full simulation in a browser to check response timing, voice quality, and how the AI adapts to unexpected answers.

Post-call, the dashboard tracks failed rate, answer rate, and median call duration, with trend views across 7, 30, and 90 days. Follow-up training triggers automatically on failure. A cooling period prevents the same scenario from targeting the same employee in rapid succession.

Brightside is available on the Pro plan or as a Voice add-on (10+ seats). Deepfake video simulation is a managed service for executive-level engagements, not a self-serve module.

Best for: Security teams that want deep self-serve control over live adaptive AI vishing calls, hybrid voice-plus-email attack scenarios, and custom voice cloning in one platform.

Pros

• AI-recommended attack strategy with psychological reasoning — unique in the market
• AI-generated caller persona and opening message auto-filled from the attack goal
• Hybrid attack workflow: coordinated call + phishing email in one campaign
• Preview-before-launch browser testing
• Custom voice cloning from a short recording
• Vishing-specific metrics dashboard (failed rate, answer rate, call duration)
• Cooling period and automatic follow-up training on failure
• Multilingual voice library expandable on request

Cons

• Vishing requires Pro plan or Voice add-on; not available on the free Start tier
• Deepfake video is a managed/white-glove engagement, not self-serve

Adaptive Security

Adaptive Security is an AI-threat-focused human risk platform aimed at enterprise security teams. Its positioning centers on automating security awareness and simulation at scale, with particular emphasis on exposure to AI-generated threats including voice-based attacks.

The platform offers simulated phishing, phone-based social engineering, and deepfake scenarios, with an administrative layer that surfaces individual employee risk scores and tracks exposure patterns across the organization. Its executive exposure monitoring makes it relevant for financial-sector teams concerned about targeted campaigns against named individuals, which is precisely the threat pattern the hedge-fund wave illustrated.

Adaptive Security is best evaluated by teams that want simulation to plug into a broader human risk posture program rather than as a standalone phone-channel tool.

Best for: Enterprise teams that want automated multi-vector simulation integrated with human risk scoring and executive exposure monitoring.

Pros

• AI-driven simulation across phishing, vishing, and deepfake scenarios
• Executive exposure monitoring relevant to high-value-target risk
• Human risk scoring at the individual level
• Designed for enterprise-scale automation

Cons

• Less granular scenario-design control for vishing than specialist simulators
• Custom voice cloning and hybrid attack capabilities not as explicitly documented as in simulation-specialist platforms
• Validate specific vishing call mechanics in a demo before committing

Arsen

Arsen is a Paris-based security awareness platform with a strong focus on multi-channel simulation, including live AI vishing calls. It covers phishing, smishing, and vishing with a simulation-first approach, and has built recognition in the European market around attack realism and scenario control.

The platform supports live adaptive AI vishing conversations (not prerecorded or voicemail-only) and includes voice cloning capability for more targeted simulations. It also offers threat monitoring features and collaboration-tool delivery that extend beyond conventional email-and-phone simulation. Arsen is worth evaluating for European teams that want a vendor with strong regional positioning and documented vishing depth.

Best for: European security-awareness teams wanting live AI vishing simulation with compliance-ready reporting and multi-channel coverage.

Pros

• Live adaptive AI vishing conversations
• Voice cloning support for personalized scenarios
• Multi-channel coverage: phishing, smishing, vishing
• Strong European market positioning and compliance framing
• Vishing-specific metrics

Cons

• Smaller platform footprint than global SAT vendors; content library may be narrower
• Hybrid voice-plus-email workflow in a single campaign is documented but worth confirming in a proof of concept
• Less established in North American enterprise buying cycles

Hoxhunt

Hoxhunt is an enterprise human risk platform with adaptive phishing simulation, gamified learning, and a threat-intelligence feed that informs simulation difficulty. It has extended into vishing and deepfake scenarios as part of a broader multi-vector offering.

The platform includes live adaptive vishing conversations and built-in deepfake voice cloning for targeted simulations. Its simulation realism in the email channel is informed by real threat-intelligence signals, and a similar principle applies to its voice scenarios. For organizations that have matured their email-phishing simulation and want to add voice coverage to an established program, Hoxhunt provides a natural extension without switching platforms.

Its combined email and meeting-based attack scenarios (Teams/Meet/Zoom plus a fake meeting invitation) add a realistic multi-channel dimension, though the vishing component is browser-based rather than a live outbound call in the traditional sense.

Best for: Enterprises with mature phishing programs extending into voice simulation within an existing Hoxhunt deployment.

Pros

• Live adaptive AI vishing as part of a multi-vector platform
• Built-in deepfake voice cloning
• Threat-intelligence-fed phishing simulation with adaptive difficulty
• SOC workflow integration for remediation
• Simulation cooling period

Cons

• Vishing implementation uses browser-based calling, not a conventional outbound phone call; confirm this fits your threat model
• Tighter value proposition for teams that don’t already use Hoxhunt for email simulation
• Custom scenario control may be less granular than specialist simulators

Jericho Security

Jericho Security is an AI-native simulation vendor with a focus on personalized multi-channel attacks. It covers phishing, vishing, and SMS, and is one of the few platforms in this comparison that offers self-serve deepfake video simulation alongside its voice simulation capability.

The platform emphasizes AI-generated pretexts and rapid scenario generation, with live adaptive AI conversations in its vishing module. It is well-funded and has been building out its enterprise capabilities. Teams evaluating Jericho should confirm the maturity of its administrative controls, reporting depth, and cooling-period implementation relative to their program requirements.

Best for: Teams that need deepfake video simulation alongside live AI vishing in a single AI-native platform.

Pros

• Live adaptive AI vishing conversations
• Self-serve deepfake video simulation — rare in the market
• AI-generated phishing pretexts with rapid generation
• Voice cloning capability
• AI-native architecture designed for modern attack scenarios

Cons

• Newer vendor; enterprise operational maturity (admin controls, audit logs, cooling periods) should be validated in a proof of concept
• Reporting depth and integration with SIEM/SOAR or identity providers may lag larger incumbents
• Pricing and packaging are less publicly documented than established platforms

Keepnet Labs

Keepnet Labs is a broad human risk management platform covering awareness training, phishing, smishing, vishing, callback phishing, and incident response tooling. Its vishing simulation module includes a voice library and is positioned as part of a multi-channel simulation suite for compliance-oriented teams.

The platform’s vishing capability uses AI text-to-speech with template-based scenarios rather than a fully generative real-time AI conversation. For teams that need to cover the vishing channel in their compliance checklist and prefer a managed, consistent scenario set over fully adaptive live AI calls, Keepnet Labs is a credible option. It also includes vishing-specific metrics alongside its broader reporting.

Best for: Compliance-oriented teams wanting broad multi-channel simulation coverage including vishing, callback phishing, and incident-response workflow in one platform.

Pros

• Broad multi-channel coverage: phishing, smishing, vishing, callback phishing, and more
• Voice cloning support for executive impersonation scenarios
• Vishing-specific metrics dashboard
• Incident response and reporting tools alongside simulation
• Strong compliance and audit reporting

Cons

• Vishing uses template-based AI text-to-speech rather than a fully generative live AI conversation; confirm whether the realism level fits your training goals
• Platform breadth can mean less simulation-design depth than specialist vishing tools
• Admin action audit log is available but worth verifying coverage in a trial

KnowBe4

KnowBe4 is the largest security awareness training vendor by market adoption, with a phishing template library, the AIDA AI automation layer, and an extensive content catalog. Its vishing simulation capability is available through the Virtual Security Training (VST) module, which supports voicemail-based simulations, and through Callback Phishing, which combines an email lure with an inbound call scenario. Both are available on higher platform tiers (Gold for VST voicemail; Diamond for Callback Phishing).

KnowBe4’s vishing capability is best understood as an awareness-building layer rather than a live adaptive AI simulation. The calls are scripted, prerecorded, or inbound-triggered rather than outbound generative AI conversations. For organizations already on the KnowBe4 platform that want to introduce employees to the idea of voice-based attacks before investing in a deeper simulation program, the existing vishing modules extend the platform meaningfully. For teams that need live AI vishing simulations that adapt in real time, a specialist platform will be more appropriate.

Best for: Organizations already using KnowBe4 for phishing simulation that want to add basic vishing awareness without changing platforms.

Pros

• Largest phishing template library and content catalog in the market
• VST voicemail simulations available on Gold tier and above
• Callback Phishing scenario available on Diamond tier
• AIDA automation layer for personalized campaign management
• Broad language coverage and mature enterprise integrations

Cons

• Vishing is not a live adaptive AI conversation: voicemail-based (VST) or inbound-triggered (Callback Phishing), depending on tier
• The most relevant vishing features are tier-gated; confirm availability on your current contract
• No custom voice cloning documented
• Live outbound AI vishing calls require moving to a specialist platform

Phished

Phished is a European security awareness platform built around automated phishing simulation with a strong emphasis on administrative automation, behavioral risk scoring, and safe-click containment. It has extended its multi-channel simulation offering to include vishing through a template library.

Phished’s vishing capability is template-based rather than live generative AI, which makes it suitable for organizations at an earlier stage of vishing program maturity that want to introduce phone-channel scenarios alongside their existing email simulation. Its automated simulation cadence and behavioral risk scoring make it operationally efficient for teams without dedicated simulation program staff. The simulation cooling period helps protect employees from over-targeting.

Best for: Teams prioritizing administrative automation and behavioral risk scoring across phishing channels, extending basic vishing awareness coverage through template-based scenarios.

Pros

• Automated simulation across phishing and vishing channels
• Behavioral risk scoring at the individual level
• Safe-click containment for phishing links
• Simulation cooling period
• Automated follow-up training on failure
• Strong European compliance and GDPR positioning

Cons

• Vishing is template-based, not a live adaptive AI conversation
• Less scenario-design control than specialist vishing simulators
• Vishing module depth is not the platform’s primary differentiation
• Custom voice cloning not documented

Proofpoint

Proofpoint is an enterprise security platform whose Security Awareness Training (SAT) module sits within a broader product ecosystem spanning email security, threat intelligence, and data loss prevention. Its awareness platform includes phishing simulation, adaptive learning paths, and a People Risk Explorer that identifies very attacked people based on real threat data.

Proofpoint’s vishing coverage includes educational content and some simulation templates, but live adaptive outbound AI vishing calls are not a core documented feature of the platform. Its strongest value for vishing-adjacent use cases is integration: threat intelligence from email security informs which employees are being targeted in the wild, which can guide where you prioritize behavioral training. For enterprises already in the Proofpoint ecosystem looking for a vendor to handle both technical controls and awareness training in one relationship, the SAT module is a logical starting point, but dedicated vishing simulation capabilities should be validated in a demo before assuming equivalence with specialist platforms.

Best for: Enterprises already using Proofpoint email security that want integrated threat-intelligence-driven awareness training without adding a second vendor for the awareness layer.

Pros

• People Risk Explorer uses real-world threat data to surface high-risk employees
• Deep integration with Proofpoint email security and threat intelligence
• Adaptive learning paths and broad phishing template library
• Thousands of templates across 42+ languages
• Strong reporting and enterprise integrations

Cons

• Live adaptive outbound AI vishing simulation is not a core documented feature; confirm vishing call capabilities in a demo
• Custom voice cloning not documented
• Vishing depth requires verification; do not assume parity with specialist simulation platforms
• Strongest value for teams already committed to the Proofpoint ecosystem

SoSafe

SoSafe is a European security awareness platform with behavior-science-driven training, phishing simulation, and a multi-channel simulation offering that includes vishing. It has strong compliance positioning for NIS2, DORA, and ISO 27001, and a regional footprint that makes it a natural choice for organizations with a European-primary compliance requirement.

SoSafe’s vishing simulation supports self-serve scenarios with role-based targeting, and its platform includes deepfake voice cloning as a simulation capability. The vishing implementation is not documented as a live outbound generative AI call in the conventional sense, so teams with advanced scenario requirements should confirm the call mechanism and real-time adaptability in a proof of concept. Its strength is the combination of behavior-science training methodology, compliance reporting, and multi-channel simulation in one coherent platform.

Best for: European enterprises with NIS2, DORA, or ISO 27001 compliance requirements wanting behavior-science-led awareness training with vishing and deepfake voice simulation.

Pros

• Vishing simulation with role-based targeting
• Deepfake voice cloning support
• Behavior-science training methodology with strong engagement and completion rates
• Detailed compliance reporting for NIS2, DORA, ISO 27001
• External risk-signal ingestion and manager-facing workflows
• Strong European regional scale and language coverage

Cons

• Vishing implementation is not documented as live outbound generative AI; confirm call mechanics and adaptability in a proof of concept
• Custom scenario design control may be less granular than simulation-specialist platforms
• Better positioned for organizations where compliance reporting is a primary driver alongside simulation

Vishing FAQs

What makes AI vishing different from traditional voice phishing?

Traditional vishing relied on human callers working from a script. Skill, preparation time, and the number of available operators capped how many targets a campaign could reach. AI vishing removes those constraints: voice synthesis generates convincing audio from a short recording, automated persona research builds a plausible pretext, and the call itself can be run by an AI agent that responds in real time. Campaigns that previously reached around 50 firms can now reach 1,000 with comparable quality. The attack hasn’t changed in kind (it’s the same social engineering), but AI has made it scalable and cheaper per call.

Does phishing-resistant MFA protect against vishing attacks?

Phishing-resistant MFA (FIDO2 security keys, passkeys) is structurally sound against credential phishing. It eliminates the “read back your OTP” and “approve this push notification” attack paths because there is no code to intercept and no push to approve under pressure.

Help-desk social engineering bypasses the MFA system by exploiting the process that manages it. An attacker calls the service desk, impersonates an enrolled employee, and requests MFA removal and re-enrollment of a new device. The MFA controls are irrelevant when the service desk removes them on the caller’s say-so, so help-desk identity proofing and out-of-band callback verification must accompany MFA.

How do help-desk vishing attacks actually work?

The attacker calls the IT service desk posing as an employee who has lost access to their device or phone. They claim they cannot receive their MFA challenge, and they ask the agent to remove the existing MFA factor and enroll a new one, usually a device the attacker controls. If the agent does not verify the caller’s identity through a source independent of the call itself (an internal photo ID system, a manager callback, or a face-to-face check for high-privilege accounts), the attacker inherits the account.

The variant used in cloud attacks like ShinyHunters and BlackFile adds a lookalike SSO portal: the vishing call directs the target to a phishing page that captures credentials and MFA codes in real time, while the attacker relays those credentials to the legitimate system. Both paths depend on the employee or help-desk agent taking an action during the call without independent verification.

What does a vishing simulation test that security awareness training does not?

Awareness training transfers knowledge: employees learn what vishing is, what red flags look like, and what they are supposed to do. What it cannot test is what employees actually do when a confident, adaptive caller is applying real-time pressure.

A vishing simulation puts the employee in that situation. It measures whether they ask for a callback number and verify it against an internal directory, whether they refuse to act on an inbound caller’s instructions and escalate instead, and how quickly they report the call even if they complied. The most effective simulation programs follow a failure with immediate, constructive remediation — not punitive scoring, but a targeted training moment delivered while the experience is fresh.

How often should organizations run vishing simulations?

A useful starting point is quarterly simulations for high-risk roles (help desk, finance, executive support) and semi-annual or annual for the broader employee population, with frequency adjusted based on prior results. New hires and employees who have recently failed a simulation warrant faster re-testing.

The goal is enough frequency to build and maintain a behavioral reflex, not so much that employees become desensitized or the program loses credibility. Most platforms support cooling periods that prevent the same scenario from repeating too quickly against the same employee, which protects both data quality and employee trust. As your program matures and failure rates fall, the structure of scenarios can increase in complexity to reflect the evolving threat.