KnowBe4 Phishing Simulation: How It Works, What It Tests, and Its Limits Against AI-Powered Phishing (2026)
Read summarized version with

Key takeaways
- KnowBe4 phishing simulation sends employees simulated phishing emails, callback phishing lures, and QR code tests, then records clicks, replies, data entry, and reports made through the KnowBe4 Phish Alert Button
- KnowBe4 phishing training pairs those tests with landing pages, remedial cybersecurity awareness training groups, and AI Defense Agents (AIDA) that select templates and assign follow-up content
- The KnowBe4 Phish-prone Percentage counts every failure against the emails delivered, so it reflects the lures chosen as much as the workforce tested
- Template-based KnowBe4 phishing tests measure inbox habits well but say little about voice, SMS, and deepfake impersonation aimed at specific roles
- An AI phishing simulation built from public information about each employee tests the personalized, multichannel lures that cyberattackers now use
- Ethical phishing simulation programs secure HR sign-off, avoid lures that exploit pay or personal hardship, and reward reporting instead of punishing clicks.
KnowBe4 phishing simulation is the Phishing Security Test feature in KnowBe4 Security Awareness Training. It sends employees simulated phishing emails, callback phishing lures, and QR code tests, records clicks, replies, data entry, and reports, and scores results as a Phish-prone Percentage. Its core strength remains template-based email testing rather than personalized, multichannel cyberattacks.

According to Gartner's 2026 Survey, 41% of 297 CISOs and equivalent cybersecurity leaders surveyed reported at least one social engineering incident involving a deepfake during an employee audio call, and 36% reported one during a video call. Many organizations still measure exposure to these cyber threats with phishing simulations that grade inbox clicks on prewritten templates, an approach that captures only one channel of a problem that now arrives by phone, text message, and video call.
KnowBe4 reports more than 70,000 customer organizations, so the design of a KnowBe4 phishing simulation shapes how a large number of security teams judge their own risk. What each KnowBe4 phishing test records, and what it leaves out, decides whether a falling score reflects safer behavior or easier lures.
This KnowBe4 phishing simulation guide covers:
- How a KnowBe4 phishing simulation campaign runs from template to report;
- Which channels a KnowBe4 phishing test covers in 2026, including KnowBe4 callback phishing and smishing;
- How the KnowBe4 Phish-prone Percentage is calculated and what it cannot show;
- Where template-based KnowBe4 phishing training falls short against AI phishing simulation;
- How to run phishing tests ethically and choose a multichannel phishing simulation platform.
Inbox-only phishing tests leave voice, text message, and deepfake lures unmeasured until a cyberattacker exploits them. Adaptive Security runs phishing simulations across every channel employees answer at work.
What Is KnowBe4 Phishing Simulation?
KnowBe4 phishing simulation is the Phishing Security Test (PST) capability inside KnowBe4 Security Awareness Training that sends controlled, simulated cyberattacks to employees and records how each one responds. KnowBe4 describes its product through three functions, Train, Test, and Reinforce, with phishing simulation as the Test function and real-time coaching as the Reinforce function.
What a KnowBe4 Phishing Test Measures
According to Verizon's 2026 Data Breach Investigations Report, a human element was present in 62% of breaches, up from 60%. A KnowBe4 phishing test measures that human element directly by logging what each employee does with a simulated phishing email. KnowBe4 documents eight employee behaviors that a phishing test can record as of October 2026:
- Opens: A small tracking image records when an employee opens the email, and administrators can switch it off;
- Link clicks: Each recipient receives a unique link, so every click is tied to one person;
- Attachment activity: Opening an attachment or enabling its macros is logged as a separate action;
- Data entry: Typing into a landing page that imitates a login form is recorded;
- Replies: Answering the simulated email is captured when reply tracking is switched on;
- QR code scans: Scanning a code in the email with a phone counts the same way as a click;
- Callback activity: Entering the code after calling the number in a callback email, and then sharing personal details, are each logged;
- Reports: Flagging the email through the KnowBe4 Phish Alert Button is recorded in campaign results.
KnowBe4 states that it never stores the text an employee types into a data-entry landing page and records only that information was submitted, provided custom landing pages follow KnowBe4's setup directions. That approach lets a KnowBe4 phishing simulation test credential-harvesting behavior without collecting real passwords.
Free KnowBe4 Phishing Simulation vs. Paid KnowBe4 Phishing Training
KnowBe4's free KnowBe4 phishing simulation tool, named the Attack Simulation Test, tests up to 100 users against one chosen vector, such as email phishing, reply phishing, social media phishing, or QR code phishing. Administrators pick a landing page that either reveals the missed red flags or shows a 404 error, and a PDF report with a Phish-prone Percentage and an industry benchmark arrives 24 hours after the test starts.
Paid KnowBe4 phishing training removes those limits through the SAT Foundation and SAT Advanced tiers, which replaced KnowBe4's earlier Silver, Gold, Platinum, and Diamond levels in May 2026. KnowBe4's feature list for the tiers includes unlimited phishing security tests year-round, recurring campaigns, and remedial cybersecurity awareness training.
KnowBe4 lists more than 200 pieces of content in the Foundation library and more than 1,000 in the Advanced library. Other tier differences include AI features and Smart Groups. One baseline test shows a snapshot, while a recurring program, as outlined in Adaptive Security's phishing simulation guide, shows whether behavior changes.
According to the FBI's 2025 IC3 Annual Report, the FBI received 191,561 phishing and spoofing complaints in 2025. Those complaints capture only incidents reported to the FBI, so a baseline test remains a practical way for an organization to measure its own exposure before committing to a recurring program.
Email-first phishing tests record inbox behavior, while text-message lures go unrecorded and deepfake calls go untested. Adaptive Security records responses across SMS, voice, and deepfake phishing simulations.
How KnowBe4 Phishing Simulation Works: Step-by-Step
A KnowBe4 phishing simulation campaign moves through six stages, from choosing KnowBe4 phishing templates to assigning remedial KnowBe4 phishing training. Each stage has settings that change what the final results mean, so the order of decisions matters as much as the templates themselves.
The six stages run in the following order:
- Template selection: Administrators pick built-in, edited, or custom templates and set a difficulty rating;
- Campaign setup: The campaign defines recipients, frequency, sending window, and tracking period;
- Delivery: KnowBe4 sends tests at random points across the sending window;
- Landing page: An employee who fails sees a page that reveals the phishing simulation;
- Reporting: Employees who spot the test report it through the KnowBe4 Phish Alert Button;
- Remediation: Employees who fail move into a group for remedial cybersecurity awareness training.
Choosing KnowBe4 Phishing Templates and Difficulty Ratings
KnowBe4 phishing templates come from a large built-in library of templates, and administrators can use them as written, edit them, or build their own. Each campaign draws only on the template topics that the administrator selects.
KnowBe4 assigns every system template a difficulty rating from one to five stars. The rating reflects grammar, spelling, punctuation, logo use, overall design, and whether the template inserts targeted details about the recipient through placeholders. KnowBe4 recommends one-star and two-star templates after employees finish introductory cybersecurity awareness training, leaving four-star and five-star templates for employees ready for messages with very few red flags.
Two settings keep a campaign unpredictable. Full Random template selection sends different employees different emails, and KnowBe4 recommends leaving the phishing link domain on its default Random Domain setting.
Automation adds a third layer of variety. AI-Selected Phishing Templates choose a template for each employee from that person's cybersecurity awareness training and phishing history, and templates generated by AI Defense Agents (AIDA) carry red-flag indicators based on the NIST Phish Scale User Guide, published as NIST Technical Note 2276 in November 2023.
Campaign Setup, Landing Pages, and the KnowBe4 Phish Alert Button
Campaign setup controls how a KnowBe4 phishing test reaches employees. Administrators choose a weekly, bi-weekly, monthly, or quarterly frequency, a start time, and a sending window that KnowBe4 advises should last at least three business days.
KnowBe4 can send every email when the campaign starts or spread delivery across that window to prevent employees from warning each other. Three further settings shape the data each campaign produces:
- Business hours: Tests arrive only during the working days and hours set for the account;
- Tracking period: KnowBe4 advises tracking activity for at least three days after sending so that late clicks still count;
- Reply tracking: Administrators switch on reply tracking for reply-to campaigns, which imitate business email compromise requests.
Landing pages determine what an employee sees after failing. KnowBe4's basic landing pages state that the email was a simulated phishing test and offer tips for inspecting future messages, while data-entry pages imitate a login form to test whether the employee types credentials. Social Engineering Indicators go further by highlighting the specific red flags the employee missed inside the original email.
The KnowBe4 Phish Alert Button is a one-click add-in that lets employees report suspicious messages from their email client, and KnowBe4 also offers it as a free tool. Reports of simulated emails appear in campaign results, which gives administrators a count of correct behavior alongside failures.
According to the UK Department for Science, Innovation and Technology's Cyber Security Breaches Survey 2025/2026, 22% of businesses tested staff, for example with mock phishing exercises. By that measure, an organization running a recurring KnowBe4 phishing simulation tests staff behavior, a step that only 22% of UK businesses in the survey reported taking.
Reporting and Remedial KnowBe4 Phishing Training
Every failure is logged against the employee who received the test, and the "Add Clickers to" setting places employees who fail into a group for a later remedial cybersecurity awareness training campaign. Administrators can also have KnowBe4 email a report to account admins after each test closes.
SAT Advanced subscribers can hand remediation to AIDA Orchestration, which KnowBe4 launched in February 2026. KnowBe4 states that the agent determines autonomously, after each failure, whether a short SecurityTip or a full KnowBe4 phishing training assignment fits the employee's risk.
Campaigns marked Hide from Reports do not affect Phish-prone Percentages or Risk Scores. KnowBe4 recommends that option for test campaigns run to confirm email allowlisting, which keeps configuration checks out of the behavioral record.
A template library means every lure starts from a script written for broad audiences. Adaptive Security's Auto-Phish builds each phishing simulation from the employee's role, public footprint, and click history.
Which Channels Does KnowBe4 Phishing Simulation Cover in 2026?

As of October 2026, a KnowBe4 phishing simulation can take eight documented forms, all delivered by email, and KnowBe4's feature list adds simulated vishing, physical QR code, and USB drive tests. Email therefore remains the entry point for nearly every KnowBe4 phishing test, including those that move the employee onto a phone call.
The following table maps each KnowBe4 phishing test type to its channel, failure trigger, and availability as of October 2026, based on KnowBe4's own documentation.
| Test type | Channel | Failure trigger | Availability |
|---|---|---|---|
| Phishing link | Clicking the link | Unlimited phishing security tests | |
| Attachment | Opening it or enabling macros | Unlimited phishing security tests | |
| Data entry | Email to a fake login page | Clicking, then entering data | Unlimited phishing security tests |
| Spear phishing | Email, custom template | Clicking or entering data | Custom build with KnowBe4 support |
| Reply-to | Replying | Phishing Reply Tracking | |
| QR code | Email, scanned by phone | Scanning the code | Documented test type |
| ClickFix | Email to a fake error page | Running a copied command | Documented test type |
| Callback phishing | Email, then phone call | Entering the code, then personal data | Support article cites Diamond level |
| Simulated vishing | Voice | Entering data via telephone keypad | SAT Advanced or Diamond |
| Physical QR code | Printed code | Scanning (details not documented) | Listed feature |
| USB drive | Physical device | Not documented | Listed feature |
| SMS (smishing) | Text message | None documented | Training content only |
Email-Based KnowBe4 Phishing Test Types
Five of KnowBe4's documented test types follow the classic email pattern:
- Phishing link: A misleading link leads to a safe landing page, and the click counts as a failure;
- Attachment: Opening the file or enabling its macros counts as a failure, and KnowBe4's built-in attachments cannot be modified or replaced with custom files;
- Data entry: A link leads to a page imitating a login page, and submitting information counts as a second failure;
- Spear phishing: A custom template targets one person or department, and KnowBe4's Customer Success Managers help build it;
- Reply-to: Answering the email counts as a failure, mirroring the reply step in business email compromise.
ClickFix works differently. The test directs the employee to a page showing a fake error message, CAPTCHA check, or software update prompt that instructs the person to paste and run a copied command, and running it counts as the failure. KnowBe4 notes that ClickFix attacks are more likely to evade detection by security tools because the malicious action is user-initiated.
KnowBe4 Callback Phishing and Simulated Vishing
According to Mandiant's M-Trends 2026, voice phishing was the initial infection vector in 11% of intrusions, the second most common, while email phishing accounted for 6%. Adaptive Security's guide to vishing vs. smishing explains how the voice and SMS channels differ.
KnowBe4 callback phishing, which KnowBe4 also calls reverse vishing, is the company's main voice-related test. The employee receives an email containing a phone number and a callback code; calling the number plays a recorded greeting, entering the code counts as one failure, and entering personal information afterward counts as a second. KnowBe4 states that it does not record the information employees enter during the call.
Callback numbers are available in more than 15 countries, with 20 regions listed as of November 2026, including the United States, the United Kingdom, and Germany. KnowBe4's callback campaign article notes that callback failures do not currently affect Risk Scores, and KnowBe4's support article still describes the feature as available to Diamond-level customers, a level KnowBe4 replaced for new purchases when the new tiers launched in May 2026.
KnowBe4's feature list also includes simulated vishing, described as a test of how employees respond to a realistic vishing call, followed by automated remedial training. KnowBe4's Vishing Product Guide documents outbound calls that play pre-recorded or text-to-speech audio, and a test fails when the employee enters the requested digits.
Is There a KnowBe4 Smishing Simulation? SMS Training, QR Code, and USB Tests
According to the Anti-Phishing Working Group's Phishing Activity Trends Report, 2nd Quarter 2026, Crane Authentication data in the report showed smishing rising 40% and vishing 20% from the first to the second quarter of 2026.
KnowBe4 does not list SMS among its phishing simulation types and makes no claim to offer a KnowBe4 smishing simulation as of October 2026. KnowBe4 covers smishing (SMS phishing) through cybersecurity awareness training content, so organizations that want to measure how employees respond to text-message lures need a provider that sends simulated SMS messages. Adaptive Security's guide to SMS phishing (smishing) explains how those lures work.
KnowBe4's remaining non-email tests are physical. A physical QR code test checks how employees react to finding an unexpected QR code, and a USB drive test checks how they react to unknown USB drives found in or near the workspace.
Where AI Fits: AIDA Templates and Deepfake Training Content
KnowBe4 announced its twelfth AI Defense Agent (AIDA) on May 5, 2026, with the full AIDA suite included in SAT Advanced and select AI features in SAT Foundation. Three agents shape a KnowBe4 phishing simulation directly:
- Template Generation Agent: Writes phishing templates whose social engineering indicators, or red flags, are based on the NIST Phish Scale;
- Callback Template Generation Agent: Writes the emails used in callback phishing tests;
- Recommended Landing Pages Agent: Suggests a landing page to pair with each generated template.
The AI shapes the lure and its landing page, while the delivery channel stays email or an email-initiated phone call. A separate agent produces deepfake training content, which sits among KnowBe4 AI features for training rather than among its phishing test types.
Measure smishing with real SMS phishing simulations instead of covering it through training content alone. Adaptive Security sends text-message lures alongside voice, deepfake, and email scenarios in one program.
How the KnowBe4 Phish-Prone Percentage Is Calculated
KnowBe4 calculates a campaign KnowBe4 Phish-prone Percentage by dividing the total number of failures by the number of emails delivered. One employee who clicks a link and then enters data counts twice, so the metric measures failures per delivered test rather than the share of employees who failed.
How KnowBe4 Counts a Failure
KnowBe4 measures unsafe actions instead of people. Every unsafe action on a test counts as one failure, so one employee can log several failures on one email.
KnowBe4 ranks the eight failure types from most to least severe:
- Entering data on a landing page;
- Entering personal information during a callback phishing call;
- Enabling a macro in an attachment;
- Replying to the simulated phishing email;
- Opening an attachment;
- Scanning a QR code;
- Calling the number and entering the code from a callback phishing email;
- Clicking the embedded link.
KnowBe4 also stops updating click data on any one test email after 10 clicks.
The KnowBe4 Phish-Prone Percentage Formula, Step by Step
The campaign KnowBe4 Phish-prone Percentage follows three steps:
- Count delivered tests: The denominator is the number of simulated phishing emails delivered in the campaign;
- Count failures: The numerator is every unsafe action recorded, using the eight failure types;
- Divide: Failures divided by delivered emails, expressed as a percentage, gives the campaign score.
KnowBe4's own worked example shows the effect. When 100 employees receive a test and 52 click the link, the campaign records 52 failures and a score of 52%. When eight of those 52 also enter data on the landing page, the campaign records 60 failures across 100 emails and a score of 60%, although only 52 people failed.
Individual scores follow the same logic. An employee who receives one test and both clicks the link and opens the attachment records two unsafe actions on one test received, which KnowBe4 displays as a 200% personal Phish-prone Percentage. KnowBe4 reads a higher personal score as a sign that the employee may be more vulnerable and tracks the exact way each user fails.
KnowBe4 also runs a companion metric, the Phish-failure Percentage, which counts each failing employee once. In the 100-employee campaign example it stays at 52%, so reporting both figures side by side shows how many people failed and how severely they failed.
What the KnowBe4 Phish-Prone Percentage Cannot Show Alone
According to UC San Diego's Understanding the Efficacy of Phishing Training in Practice 2025, average failure rates across ten simulated lures sent to 19,789 healthcare employees ranged from 1.82% for an "Outlook password" lure to 30.80% for a "vacation policy" lure.
That randomized study of phishing training at UC San Diego Health shows that lure choice can move failure rates far more than any training effect. A falling KnowBe4 Phish-prone Percentage may therefore reflect easier templates as much as safer behavior, which is why the difficulty rating of each campaign belongs next to the score in every report.
The following table compares five phishing simulation metrics by what each measures and what each cannot show alone.
| Metric | What it measures | What it cannot show alone |
|---|---|---|
| Phish-prone Percentage | Failures per delivered email | Lure realism; non-email channels |
| Phish-failure Percentage | Share of recipients who failed at least once | How severely each person failed |
| Reporting rate | How often employees report suspicious messages | Whether reports came fast enough |
| Risk Score | A per-user composite of behavior signals | Callback failures, currently excluded as of November 2026 |
| Training completion | Whether modules were finished | Whether behavior changed |
For measuring behavior under pressure, failure metrics and reporting rates together give the most direct evidence. The UK National Cyber Security Centre (NCSC) recommends counting how many employees reported a phishing email alongside how many clicked.
When a risk score leaves out phone-based failures, it understates exposure to voice phishing. Adaptive Security's employee risk scores update from phishing simulation results, training completion, and role-based exposure.
What Template-Based KnowBe4 Phishing Training Measures Well and What It Misses
Template-based KnowBe4 phishing training measures how employees treat generic inbox lures at scale and over time. It measures far less about personalized, multichannel cyberattacks, in which cyberattackers combine email with phone calls, text messages, and cloned voices aimed at specific roles.
Where Template-Based Phishing Tests Deliver Reliable Data
A template-based KnowBe4 phishing test produces consistent, repeatable data. Because the same library, difficulty ratings, and scoring rules apply to every campaign, an organization can set a baseline, test thousands of employees in their own languages, and track the trend from quarter to quarter.
Experience the Adaptive platform
Take a free tourTemplate testing also measures reporting habits well. Every report made through the KnowBe4 Phish Alert Button against a known test email is a confirmed correct action, which gives security teams a positive behavior to count alongside failures.
Difficulty ratings add further control. Because every system template carries a star rating, administrators can hold nominal difficulty steady from one campaign to the next, which narrows the effect of lure choice on results without removing it.
Administrators value that predictability. A small-business reviewer on G2 in July 2026 praised how simply KnowBe4 phishing training campaigns can be scheduled each quarter, with new starters enrolled automatically.
Where Phishing Template Libraries Fall Short Against AI Phishing

According to Verizon's 2026 Data Breach Investigations Report, the median click rate in phishing simulations was 40% higher for mobile-centric vectors, such as voice and text messaging, than for email. Template libraries built mainly for the inbox therefore test the channel where employees already perform best.
KnowBe4 phishing templates are written for broad audiences, while AI-written lures increasingly reference the target's role, vendors, travel plans, and recent public posts. A library template can insert a name and department through placeholders, yet it cannot draw on the specific context that makes a spear phishing message credible to one finance approver.
Channel switching is harder still to reproduce. Real campaigns can move from an email to a text message to a phone call, and a template-based KnowBe4 phishing simulation stages each channel separately, with callback phishing, which pairs a simulated email and phone call in one campaign, as the email-to-phone chain in KnowBe4's list of test types.
Library depth also wears thin with repetition. A mid-market G2 reviewer in August 2026 wrote that the entry package's content library felt limited over time, and repeated exposure to the same library risks teaching employees to recognize familiar tests instead of real red flags.
What Controlled Studies Show About Embedded Phishing Training
Embedded phishing training, in which an employee who fails a phishing simulation lands on a teaching page, has been tested in two large randomized studies. The UC San Diego Health study randomly assigned employees to control and training groups across eight months of campaigns between January and October 2023, while a separate study presented at the 2022 IEEE Symposium on Security and Privacy followed employees at one large company for 15 months.
According to UC San Diego's Understanding the Efficacy of Phishing Training in Practice 2025, 56% of employees clicked a simulated phishing link at least once over eight months, whether or not they had received training.
Embedded training produced a statistically significant but small effect in that healthcare study. For the annual cybersecurity awareness training the organization used, built on KnowBe4 material, the researchers found no relationship between how long ago employees completed it and how often they failed. The 2022 study found that embedded training did not make employees more resilient and could make some of them more susceptible.
Each study describes one organization, yet together they show that phishing simulation results depend heavily on lure design and on whether employees engage with the follow-up. That finding strengthens the case for realistic, role-specific scenarios that match the cyberattacks each team receives.
What happens when a phishing request moves from email to text message to phone call? Adaptive Security chains email, voice, SMS, and chat into one coordinated phishing simulation for employees.
Template-Based vs. AI-Powered Phishing Simulations
Realism is the clearest dividing line between the two approaches. Template-based testing reuses prewritten lures for many recipients, while an AI phishing simulation builds each lure from public information about the specific employee and can carry it across email, text message, and voice.
Template-Based vs. AI-Personalized Simulation by Realism
The following table compares template-based and AI-personalized phishing simulation across seven realism criteria.
| Realism criterion | Template-based phishing simulation | AI-personalized phishing simulation | What the difference changes |
|---|---|---|---|
| Lure source | Prewritten library template | Generated from the employee's public footprint and role | Whether the test resembles mail the employee expects |
| Personalization | Name, department, and placeholder fields | Role, vendors, projects, events, and colleagues | Exposure to spear phishing instead of bulk phishing |
| Channel | Mostly email, with QR code and callback variants | Email, SMS, voice, and video, alone or chained | Whether channel switching is ever practiced |
| Timing and context | Sent within a scheduled window | Tied to plausible events such as travel, renewals, and payroll dates | Pressure and urgency closer to real cyberattacks |
| Difficulty control | Star rating chosen by the administrator | Adjusted per person from past behavior | Comparability across groups versus individual challenge |
| Benchmarking | Easy to compare across organizations | Harder to compare because each lure differs | Trend tracking versus realism |
| Consent and ethics overhead | Low | Higher, because executive likeness and personal data are involved | Need for HR, legal, and executive sign-off |
Template-based testing suits baselines, trend lines, and compliance evidence. AI-personalized testing suits the roles that cyberattackers target individually, such as finance approvers, help desk staff, and executive assistants, where one convincing request can move money or reset access.
Annotated Example: An AI-Personalized Lure Built From Public Information
According to the FBI's 2025 IC3 Annual Report, businesses reported more than $30 million in losses to business email compromise scams involving AI in 2025. The FBI notes that voice cloning can be used to request wire payments, the move the following scenario models.
The following example is a hypothetical scenario involving a fictional company and fictional employees. An accounts payable specialist at a fictional manufacturer receives an email naming a supplier featured on the company's public case-study page and mentioning the chief financial officer's (CFO's) conference appearance posted on a professional networking site. A text message from someone claiming to be the CFO's assistant follows, and then a short voicemail in a cloned version of the CFO's voice asks the specialist to pay the supplier through changed bank details before a stated cutoff.
The voicemail relies on techniques covered in Adaptive Security's analysis of AI vishing and voice spoofing, and six red flags expose the lure:
- Changed payment details: A request to pay an existing supplier through new bank details is a hallmark of business email compromise and needs independent verification;
- Channel hopping: Moving from email to text message to voice spreads the request across channels so that no one message looks complete;
- Borrowed context: The supplier name and conference reference come from public sources, so their accuracy proves nothing about the sender;
- Manufactured deadline: A cutoff time pressures the employee to skip normal approval steps;
- Unavailable approver: The CFO appears reachable only through an assistant and a voicemail, which blocks a quick confirmation;
- Verification bypass: Every contact method in the lure comes from the cyberattacker, so replying through any of them confirms nothing.
A callback to the CFO on a number from the company directory defeats this lure even when the cloned voice sounds convincing, because the cyberattacker no longer controls the verification step.
How AI Phishing Simulation Changes What a Test Can Measure
An AI phishing simulation measures behavior that template tests rarely reach. It shows how an accounts payable team responds to role-specific pressure, whether employees verify requests across channels, and whether they report lures that arrive by text message or phone as readily as email.
The approach carries trade-offs. Personalized lures are harder to compare across organizations because each message differs, and building them from public information requires governance over how data about employees is collected and stored. The UK National Cyber Security Centre (NCSC) also notes that cyberattackers use information freely available on websites and social media, and personalized phishing simulations show which public details a cyberattacker could use.
According to CrowdStrike's 2026 Global Threat Report, AI-enabled adversaries increased their operations by 89% year over year. Testing employees against AI-personalized lures keeps phishing simulation results aligned with the cyberattacks those adversaries now run.
Adaptive Security can stage this exact scenario as one phishing simulation, sending an accounts payable team an email, a follow-up text message, and a deepfake voicemail from a cloned executive.
Can KnowBe4 Simulate AI-Generated or Deepfake Phishing Attacks?

As of October 2026, KnowBe4 generates AI-written phishing and callback templates through its AI Defense Agents and offers a Deepfake Training Content Agent that produces custom training content featuring an organization's own leader. KnowBe4's published materials describe deepfake training content rather than live deepfake call or video phishing simulations.
AI Phishing Simulation in KnowBe4: AIDA Template Generation
KnowBe4's version of AI phishing simulation centers on automation. The Orchestration Agent, launched in February 2026, generates and schedules phishing tests and cybersecurity awareness training for each employee, and its Phishing Agent sub-agent creates and delivers personalized phishing simulations without administrator-built campaigns.
Personalization in this model draws on each employee's phishing and cybersecurity awareness training history inside KnowBe4. The generated message still reaches the employee as an email or an email-initiated callback, so AI changes the content of a KnowBe4 phishing simulation while the channel stays the same.
For administrators, the main change is workload. The agent replaces manual campaign building, so administrators review results instead of assembling each campaign by hand.
Deepfake Training Content vs. Live Deepfake Simulation
KnowBe4's Deepfake Training Content Agent, included in SAT Advanced, generates custom training content featuring a leader from the customer's organization. Employees watch the deepfake inside a training module, which shows them what a cloned executive looks and sounds like in a controlled setting.
A live deepfake phishing simulation tests a different skill. An unannounced call or video message that uses an executive's cloned voice or likeness measures whether the employee pauses and verifies before acting, which is the decision that matters during a real cyberattack.
The distinction also changes what can be measured. KnowBe4 tracks training by completion and quiz scores through its Track Scores option, while phishing link clicks are recorded as failures.
Organizations evaluating KnowBe4 deepfake training should treat the agent's output as awareness content. Buyers who need live deepfake testing should ask in a product demo whether KnowBe4 offers it or plans to.
Why Deepfake Coverage Matters for Finance and Help Desk Teams
According to IBM's Cost of a Data Breach Report 2026, one in four malicious breaches were AI-enabled, a 56% increase, and they averaged $6 million.
Finance approvers and help desk agents sit at the decision points those cyberattacks target, because one approved payment or one password reset can hand a cyberattacker money or access. Voice and video impersonation aimed at these roles exploits trust in a familiar face or voice instead of a suspicious link. Help desk staff face the same pattern when a caller impersonates an employee to request a password or multifactor authentication reset.
The UK National Cyber Security Centre (NCSC) advises verifying important requests through a second communication channel, such as a phone call. A deepfake phishing simulation is the most direct way to see whether employees in these roles follow that rule under pressure.
If an employee has seen deepfakes only in a training video, the organization cannot predict a live response. Adaptive Security tests that decision with realistic, customizable executive deepfake simulations.
Running a KnowBe4 Phishing Test Ethically
An ethical KnowBe4 phishing test has HR and legal sign-off, avoids lures that exploit pay, health, or personal hardship, and treats failures as coaching moments. Any phishing simulation that uses an executive's voice or likeness also needs that executive's explicit consent before it runs.
Get HR and Legal Sign-Off Before Any Phishing Simulation
The NCSC guidance on defending against phishing states that no training package, including phishing simulations, can teach employees to spot every phishing email. The guidance warns that punishing employees for clicking on emails an organization sent itself starts to resemble entrapment, and it advises checking with HR before running any phishing simulation.
A written approval record should cover three points:
- Purpose: what the program measures and how the results will be used,
- Data: which actions are recorded, such as clicks, replies, and call activity,
- Access: who can see individual results and how long they are kept.
Documenting those points before the first KnowBe4 phishing test gives HR and legal teams a clear basis for answering employee questions later.
Lures That Cross the Line: The 2020 Holiday Bonus Test
On December 14, 2020, GoDaddy sent employees an email promising a $650 one-time holiday bonus in place of the annual holiday party, and the message asked recipients to provide details to claim it. The Copper Courier reported that the email was a holiday bonus phishing test run by the company itself.
Roughly 500 employees clicked, and two days later the company's security leadership told them they had failed and needed to retake its security awareness social engineering training. In a statement reported by Engadget on December 24, 2020, a GoDaddy spokesperson said the company had apologized to employees who felt the test was insensitive.
Themes built on pay, benefits, layoffs, or health damage trust faster than they measure risk, which makes them a poor fit for any KnowBe4 phishing test. Lures can still mimic real cyberattacks through invoices, shared documents, and account alerts without exploiting employees' personal circumstances.
Consent Rules for Executive Voice and Deepfake Simulations
Deepfake phishing simulations raise a separate consent question because they reproduce a real person's voice or face. Before any executive's likeness appears in a test, that executive should give written consent, approve the script, and agree on how voice samples and video are stored and deleted.
Employees also deserve notice at the policy level. An announcement that the program includes phishing simulations by email, text message, and phone, without revealing timing or content, keeps tests realistic while avoiding the sense of ambush that damages trust.
Security teams should keep a register of every approved persona so that a cloned voice is never reused outside the scenarios its owner approved. The register also gives auditors one record of who agreed to what and when.
Reward Reporting Rather Than Punishing Clicks
The NCSC advises against reprimanding employees who struggle to recognize phishing emails, because employees who fear reprisals report mistakes late or not at all.
According to UC San Diego's Understanding the Efficacy of Phishing Training in Practice 2025, more than half of embedded training sessions ended within 10 seconds and fewer than 24% of employees formally completed the training.
Punitive follow-up is unlikely to close that engagement gap. Fast feedback on every report, and visible recognition for employees who escalate suspicious messages, give a KnowBe4 phishing test program a behavior that employees want to repeat.
Do employees get to see what happened to the message they reported? Adaptive Security's Phish Triage gives every reporter immediate feedback on the verdict, which acknowledges their contribution and makes reporting worth it.
Which Phishing Simulation Platforms Go Beyond Email?

A multichannel phishing simulation platform delivers and scores tests by SMS, voice, and deepfake video, chains those channels together, and personalizes lures for each role. Buyers should confirm each channel in a live demo rather than rely on a feature list, since listed capabilities and delivered tests can differ.
Six Criteria for Comparing Multichannel Phishing Simulation Platforms
According to ENISA's ENISA Threat Landscape 2025, phishing, including vishing, malspam, and malvertising, was the dominant intrusion vector at 60%. Evaluating a multichannel phishing simulation platform against that full range of channels calls for six criteria:
- Channel coverage: The platform sends and scores phishing simulations by email, SMS, voice, and video, and a demo should show each channel reaching a real test device;
- Channel chaining: One scenario can move from email to text message to phone, which tests whether employees verify when a request follows them across channels;
- Personalization from public information: Lures draw on each employee's role and public footprint, and buyers should ask how that data is gathered, stored, and deleted;
- Executive likeness with consent controls: The platform can clone an approved executive's voice or face for a phishing simulation and records that executive's consent;
- Metrics beyond click rate: Reports show failure severity, reporting rates, and time to report alongside clicks, broken down by role and department;
- Remediation and reporting loop: Failures trigger targeted cybersecurity awareness training automatically, and reported messages feed back into new scenarios.
These criteria mirror the questions in Adaptive Security's phishing simulation tools compared guide, which applies them across the wider market.
How KnowBe4 Phishing Simulation Measures Against Those Criteria
According to the same 2026 Gartner survey, 58% of CISOs reported a vishing or smishing incident. Against that pressure, a KnowBe4 phishing simulation program covers some channels with tests and leaves others to training content.
The following table rates KnowBe4 phishing simulation against the six criteria as of October 2026.
| Criterion | KnowBe4 status |
|---|---|
| Channel coverage | Email test types, callback phishing, simulated vishing, and physical QR code and USB tests; smishing covered by training content, with no smishing simulation |
| Channel chaining | Callback phishing chains email to phone; no other chaining documented |
| Personalization | AI-selected and AI-generated templates based on each user's history; custom spear phishing templates through a Customer Success Manager |
| Executive likeness | Deepfake Training Content Agent for training content |
| Metrics beyond click rate | Phish-prone Percentage, Phish-failure Percentage, Risk Score, reporting, and more than 60 built-in reports |
| Remediation | Remedial groups; AIDA Orchestration in SAT Advanced |
KnowBe4's May 2026 list prices put SAT Foundation at $2.40 and SAT Advanced at $3.75 per seat per month for 25 to 50 seats on three-year terms, with lower per-seat prices in larger seat bands. A side-by-side breakdown of both products appears in the Adaptive Security vs. KnowBe4 comparison.
Feature lists promise channel coverage, while SMS and live deepfake tests put it under pressure. Adaptive Security's self-guided tour shows email, SMS, voice, and deepfake phishing simulations in one product.
Multichannel Phishing Simulation for AI-Era Cyberattacks With Adaptive Security

Security teams gain evidence of how employees respond to the channels cyberattackers now use, including text messages, phone calls, deepfake video, and chained campaigns. Adaptive Security delivers that evidence through multichannel phishing simulations, an AI phishing simulation engine that generates email, SMS, voice, and deepfake scenarios from each employee's role and public footprint and chains them the way malicious phishing campaigns do.
Every multichannel phishing simulation result then feeds one risk view in Risk Monitoring and Mitigation, which gives each employee a daily score on a 100-point scale built from phishing simulation performance, cybersecurity awareness training completion, and role-based exposure.
Scores roll up by department and location, and open-source intelligence (OSINT) profiles show what cyberattackers can learn about executives and other high-value targets. When a score crosses a set threshold, Adaptive Security can enroll the employee in targeted cybersecurity awareness training automatically.
Reporting finishes the loop. Phish Triage adds a Phish Alert Button for Gmail, Outlook, and mobile, classifies every reported email as safe, spam, or malicious with AI, and removes malicious matches across the organization, with every action reversible and logged. Its Phish Remix feature turns real lures that employees report into new phishing simulation scenarios, so new scenarios track the lures employees actually receive.
See a complete multichannel phishing simulation program in action. Adaptive Security feeds email, SMS, voice, and deepfake results and reported phishing into the same employee risk scores.
Frequently Asked Questions About KnowBe4 Phishing Simulation
How Often Should a KnowBe4 Phishing Test Run?
A KnowBe4 phishing test should run at least monthly, with more frequent tests for high-risk roles. KnowBe4 itself recommends phishing all users at least once a month and high-risk users every two weeks. Varying the timing, templates, and channels from one test to the next keeps employees alert to real lures instead of to the testing schedule.
Do Employees Know a KnowBe4 Phishing Simulation Is a Test?
Employees should learn that a KnowBe4 phishing simulation program exists through policy, while the timing and content of each test stay undisclosed. KnowBe4 sends each email at a random point within the campaign's sending window so colleagues cannot warn each other, and its landing pages reveal the phishing simulation only after an employee fails.
How Should Organizations Handle Repeat Clickers in KnowBe4 Phishing Training?
Organizations should place repeat clickers in a Smart Group or remedial group and assign short, interactive KnowBe4 phishing training tied to the red flags they missed. Security teams should also review whether the employee's role needs process controls, such as callback verification for payment changes, because a control protects the organization while behavior catches up. Escalating penalties tends to suppress reporting.
Is KnowBe4 Phishing Training Free?
KnowBe4 phishing training is a paid subscription, although two KnowBe4 tools are free. The Attack Simulation Test runs a one-time phishing test for up to 100 users, and the Phish Alert Button add-in is free to install. Ongoing phishing tests and the full training library require a paid SAT Foundation or SAT Advanced subscription as of October 2026.
Is KnowBe4 Phishing Training AI-Generated?
KnowBe4 phishing training is partly AI-generated as of 2026. Most training content comes from the KnowBe4 Studios library, while AI Defense Agents select and generate phishing templates. The Content Creation Agent, announced on May 5, 2026, builds custom multilingual training modules from prompts or an organization's internal policies.
Should Employees Be Disciplined for Failing a KnowBe4 Phishing Test?
Discipline for failing a KnowBe4 phishing test depends on each organization's policy, and punishment tends to backfire. Employees who fear reprisals report mistakes late or not at all, and punishing clicks on emails the organization sent itself can resemble entrapment. Coaching, process controls, and HR-approved policies produce better outcomes, and organizations with legal questions should consult employment counsel.
Why Do KnowBe4 Phishing Tests Record Clicks Employees Never Made?
KnowBe4 phishing tests can record clicks that employees never made for two common reasons. Email security tools that open links automatically can trigger clicks, and KnowBe4 records any click on a forwarded test email against the original recipient. KnowBe4 recommends allowlisting its hostnames or IP addresses in email filtering services to remove false failures from test results.
Can a KnowBe4 Phishing Test Run in Languages Other Than English?
A KnowBe4 phishing test can run in many languages. The free Attack Simulation Test lets administrators choose the phishing language, and paid campaigns can override the account's default phishing language with up to five languages per campaign, so multinational workforces receive tests in the languages they read at work.
A clean score on email tests can hide untested chat, voice, and deepfake exposure across the workforce. Adaptive Security measures every channel with AI phishing simulations and automated follow-up.
As experts in cybersecurity insights and AI threat analysis, the Adaptive Security Team is sharing its expertise with organizations.
Get started with Adaptive Security
Related articles

Kevin Mitnick Security Awareness Training: What It Covers, How Long It Takes, and How It Compares in 2026

Enterprise Cybersecurity Awareness Training Providers: 2026 Guide to Comparing Platforms, Risk Measurement, and Enterprise Fit

