Social Engineering Deep Dive — The Human Attack Surface
Social Engineering Deep Dive — The Human Attack Surface
CIPHER Training Module | Classification: Authorized Security Training Sources: SEORG Framework, MITRE ATT&CK, TrustedSec SET, GoPhish, OWASP Last Updated: 2026-03-14
Table of Contents
- Psychological Foundations: Cialdini's Principles
- Pretexting Development Methodology
- Elicitation Techniques
- Vishing — Voice Phishing
- Smishing — SMS Phishing
- Email Phishing Taxonomy
- Physical Social Engineering
- Watering Hole Attacks
- Callback Phishing (BazarCall)
- QR Code Phishing (Quishing)
- Deepfake Threats in Social Engineering
- Tooling and Frameworks
- MITRE ATT&CK Mapping
- Defense: Security Awareness Program Design
1. Psychological Foundations: Cialdini's Principles
Social engineering exploits human psychology rather than technical vulnerabilities. The Social-Engineer.org Framework defines influence as "the process of getting someone else to want to do, react, think, or believe in the way you want them to." [CONFIRMED]
1.1 The 6+2 Principles of Influence
Robert Cialdini's original six principles, plus two additions recognized in the SE community:
1. Reciprocity
Principle: People feel obligated to return favors. A gift or concession creates psychological debt.
SE Application:
- Attacker provides "free" security audit results, then requests network access to "verify findings"
- Visher offers to "help fix" a fabricated IT issue, then requests credentials
- Phishing email delivers a "free report" requiring login to download
Why it works: The reciprocity norm is deeply embedded across cultures. Refusing a return favor creates cognitive dissonance. [CONFIRMED — Cialdini, 2001]
2. Commitment and Consistency
Principle: Once people commit to something (even trivially), they behave consistently with that commitment.
SE Application:
- Start with small, innocuous requests ("Can you confirm your department?") then escalate ("And your employee ID for verification?")
- Get target to agree they value security, then leverage that commitment to extract "verification" information
- Foot-in-the-door technique: small initial compliance leads to larger requests
Why it works: Inconsistency is psychologically uncomfortable. Once a position is taken, people rationalize continued compliance. [CONFIRMED]
3. Social Proof
Principle: People look to others' behavior to determine correct action, especially under uncertainty.
SE Application:
- "Everyone in your department has already completed this verification"
- Phishing landing pages showing "1,247 employees have already updated their credentials"
- Impersonator in a building: "Sarah from accounting let me in last time"
- Fake reviews/testimonials on credential harvesting sites
Why it works: Uncertainty + perceived peer behavior = compliance. Strongest when the "peers" are similar to the target. [CONFIRMED]
4. Authority
Principle: People defer to perceived experts and authority figures without independent verification.
SE Application:
- Impersonating IT administrators, executives, auditors, law enforcement
- Spoofed emails from "CEO" requesting urgent wire transfers (whaling/BEC)
- Wearing uniforms, badges, carrying clipboards during physical SE
- Using technical jargon to establish expertise credibility
- Vishing as "bank fraud department" or "IRS agent"
Why it works: Milgram's obedience experiments demonstrated that ~65% of people will comply with authority directives even against their judgment. [CONFIRMED — Milgram, 1963]
5. Liking
Principle: People are more easily influenced by those they like — based on similarity, compliments, familiarity, or physical attractiveness.
SE Application:
- Building rapport through shared interests discovered via OSINT (LinkedIn, social media)
- Mirroring communication style, vocabulary, and body language
- Complimenting the target's expertise: "I heard you're the go-to person for this"
- Using humor and warmth to lower defenses
- Name-dropping mutual connections
Why it works: The halo effect — positive feelings toward a person reduce critical evaluation of their requests. [CONFIRMED]
6. Scarcity
Principle: Perceived scarcity increases perceived value. Loss aversion is stronger than gain motivation.
SE Application:
- "Your account will be locked in 24 hours if you don't verify"
- "This security update expires at midnight"
- "Only the first 50 employees get the new laptop model — confirm your details now"
- Phishing emails with countdown timers
- Vishing: "I can only hold this refund for the next 10 minutes"
Why it works: Loss aversion (Kahneman & Tversky) — people fear losing something ~2x more than they value gaining something equivalent. Urgency bypasses deliberate thought. [CONFIRMED]
7. Obligation (Extended Principle)
Principle: Creating a sense of duty or indebtedness that compels action.
SE Application:
- "As part of your compliance requirements, you need to complete this form"
- Framing requests as organizational obligations
- Leveraging existing relationships: "Your manager asked me to reach out to you about this"
- Creating artificial obligations through prior "favors"
Distinction from reciprocity: Obligation focuses on duty/role expectations rather than personal favor exchange. [CONFIRMED — SEORG Framework]
8. Concession
Principle: When someone makes a concession (retreats from a larger request), the other party feels compelled to reciprocate with a concession of their own.
SE Application:
- Door-in-the-face technique: request full database access, "settle" for read-only credentials
- "I originally needed to audit the entire server room, but just let me check this one rack"
- Asking for SSN, then "accepting" just date of birth and employee ID as alternatives
Why it works: The retreat appears reasonable, and refusing a "reasonable" compromise feels socially unacceptable. [CONFIRMED]
1.2 Additional Psychological Levers
| Lever | Mechanism | SE Use |
|---|---|---|
| Fear/Threat | Amygdala hijack bypasses rational processing | "Your system is compromised right now" |
| Curiosity | Information gap theory drives completion seeking | "See who viewed your profile" / USB baiting |
| Greed | Reward seeking overrides caution | Prize notifications, fake job offers |
| Helpfulness | Desire to assist is socially reinforced | Help desk exploitation, "I'm locked out" |
| Fatigue | Decision fatigue degrades critical thinking | Late-day/end-of-week attack timing |
| Distraction | Cognitive load reduces vigilance | Multi-step processes, time pressure + complexity |
| Trust transfer | Trust in one entity transfers to associated entities | Compromised vendor emails, supply chain pretexts |
2. Pretexting Development Methodology
The SEORG Framework defines pretexting as "presenting oneself as someone else in order to obtain private information." It can extend to creating entirely new identities to manipulate information disclosure. [CONFIRMED]
2.1 Research Phase (OSINT Foundation)
Principle: "Good information gathering techniques can make or break a good pretext." Mimicking a tech support representative fails if the organization doesn't use external support. [CONFIRMED — SEORG]
Target Intelligence Collection
| Source | Intelligence Gathered | Tools |
|---|---|---|
| Org chart, job titles, reporting structure, technology stack, recent hires | LinkedIn Recruiter, PhantomBuster | |
| Company website | Products, services, partners, leadership, press releases | HTTrack, Wayback Machine |
| Job postings | Internal tools, security products, infrastructure details | Indeed, Glassdoor |
| Social media | Personal interests, travel, life events, communication style | Maltego, SpiderFoot |
| SEC/Public filings | Financials, acquisitions, key personnel, vendors | EDGAR, OpenCorporates |
| Data breaches | Email formats, credentials, internal docs | HaveIBeenPwned, DeHashed |
| DNS/WHOIS | Infrastructure, email servers, hosting | Amass, Subfinder |
| Physical recon | Building layout, badge types, visitor procedures, dumpster contents | Camera, binoculars, patience |
Organizational Intelligence Map
TARGET ORG
├── People
│ ├── C-Suite (names, assistants, email format)
│ ├── IT Staff (helpdesk numbers, ticketing system)
│ ├── HR/Finance (payroll system, benefits provider)
│ └── Target employees (role, tenure, interests)
├── Technology
│ ├── Email platform (O365, Google Workspace, Exchange)
│ ├── VPN/Remote access solution
│ ├── Helpdesk/ticketing system
│ └── Security tools (EDR, email gateway, MFA type)
├── Processes
│ ├── Visitor policy
│ ├── Password reset procedure
│ ├── Wire transfer approval chain
│ └── Vendor onboarding process
└── Culture
├── Communication style (formal/informal)
├── Dress code
├── Work hours and shift patterns
└── Recent events (layoffs, mergers, moves)
2.2 Persona Creation
Persona Development Checklist
- Identity foundation: Name, role, organization, contact info (burner phone, spoofed email)
- Backstory: Employment history, reason for contact, relevant technical knowledge
- Motivation: Why is this person contacting the target? What is natural and expected?
- Knowledge boundaries: What should the persona know? What shouldn't they know?
- Emotional state: Rushed executive? Friendly vendor? Frustrated customer?
- Verification resilience: Can the persona withstand callback verification? Supervisor check?
Common Effective Pretexts
| Pretext | Authority Level | Verification Risk | Typical Target |
|---|---|---|---|
| IT Support / Helpdesk | Medium | Medium — may call IT | End users |
| External Auditor | High | High — may verify with management | Finance, IT |
| New Employee | Low | Low — expected to ask questions | Helpdesk, HR |
| Vendor/Contractor | Medium | Medium — may check vendor list | Facilities, IT |
| Executive Assistant | High | Medium — may verify with exec | Finance, HR |
| Delivery Person | Low | Very Low | Reception, facilities |
| Building Inspector | High | Low — rarely verified | Facilities |
| Recruiter | Medium | Low | Employees (off-guard) |
2.3 Story Development
The Key Principle: Like "inserting the proper key in a lock," an effective pretext precisely matches situational expectations. [CONFIRMED — SEORG]
Story Architecture
HOOK — Why am I contacting you? (natural, expected reason)
BUILDUP — Establish credibility through knowledge demonstration
ASK — The information or access request (embedded naturally)
URGENCY — Time pressure or consequence that prevents deliberation
EXIT — Clean disengagement that doesn't trigger retrospective suspicion
Story Validation Checks
- Does this pretext make sense for this organization at this time?
- Would this person realistically contact this target through this channel?
- Can the persona answer follow-up questions without breaking character?
- Is the request proportional to the relationship established?
- Does the urgency feel natural, not artificial?
2.4 Props and Supporting Materials
Physical SE props:
- Branded polo shirts, lanyards, badges (can be created from photos)
- Clipboards, work orders, printed emails
- Toolbelts, hard hats, safety vests (instant authority for facility access)
- Branded vehicle magnets
- Business cards (trivially printed)
Digital SE props:
- Spoofed email domains (typosquatting:
cornpany.comvscompany.com) - Cloned login pages
- Fake calendar invites
- Spoofed caller ID
- Professional-looking PDF reports or invoices
3. Elicitation Techniques
The SEORG Framework defines elicitation as "the act of obtaining information without directly asking for it." The NSA characterizes it similarly as extracting information from someone or something. References include the CIA's KUBARK interrogation manual. [CONFIRMED — SEORG]
3.1 Core Elicitation Tactics
| Technique | Mechanism | Example |
|---|---|---|
| Deliberate False Statement | People correct wrong information reflexively | "I heard you're still running Windows Server 2012?" (target corrects: "No, we migrated to 2019 last year") |
| Flattery | Ego gratification loosens information control | "You clearly know more about this system than anyone — how does the authentication actually work?" |
| Mutual Interest | Shared passion creates trust and openness | Discussing industry conferences, then pivoting to internal tool discussions |
| Assumed Knowledge | Acting as if you already know triggers confirmation | "So the VPN termination is on the Palo Alto, right?" |
| Bracketing | Providing a range forces correction to the real value | "I'd guess you have somewhere between 500 and 5000 employees?" → "Actually, about 1,200" |
| Naivete | Appearing ignorant encourages explanation | "I'm sorry, I'm new to this — how does the badge access system work exactly?" |
| Quid Pro Quo | Trading information creates reciprocal disclosure | Sharing (fake) internal details to encourage target to share theirs |
| Oblique Reference | Indirect mention prompts voluntary elaboration | "Things must be crazy with the migration..." → target volunteers details |
3.2 Preloading
Preloading involves seeding ideas or framing before the actual elicitation attempt:
- Priming: Mentioning related concepts before asking (discussing "security breaches in the news" before asking about the target's security posture)
- Anchoring: Establishing a reference point that biases subsequent responses
- Context setting: Creating a conversational frame that makes disclosure feel natural
- Emotional preloading: Establishing an emotional state (anxiety, trust, urgency) that facilitates compliance
3.3 Conversation Flow Model
Phase 1: RAPPORT — Establish connection, find common ground (2-5 min)
Phase 2: TRANSITION — Natural topic bridge to area of interest
Phase 3: ELICIT — Extract information using techniques above
Phase 4: VALIDATE — Cross-reference with other questions/sources
Phase 5: EXIT — Graceful disengage without raising suspicion
Critical rule: Never ask directly for what you want. Every target question should be embedded in a natural conversational context.
4. Vishing — Voice Phishing
4.1 Definition and Impact
SEORG defines vishing as "the practice of eliciting information or attempting to influence action via the telephone." Americans lost $29.8 million to phone scams in 2021 alone. [CONFIRMED — SEORG]
4.2 Core Vishing Techniques
Caller ID Spoofing
- Services allow displaying arbitrary caller ID numbers
- Spoof the target organization's own numbers, vendor numbers, or government agencies
- VoIP services (Twilio, etc.) enable programmatic caller ID control
Voice Manipulation
- AI voice synthesis: The 2019 CEO deepfake case used AI-synthesized voice impersonating a German executive to extract $243,000 [CONFIRMED — SEORG]
- Voice changers to alter gender, age, accent
- Background noise generators (office sounds, call center ambience)
The Mumble Technique
Speaking unclearly or impersonating impaired speech to bypass verification procedures. Support staff often accommodate rather than challenge. [CONFIRMED — SEORG]
4.3 Vishing Script Templates
IT Helpdesk Pretext
"Hi, this is [Name] from IT Support. We're seeing some unusual activity
on your workstation — [workstation name from OSINT]. Your system may
have been flagged by our security monitoring. I need to verify your
identity before I can look into this further. Can you confirm your
employee ID and the email address associated with your account?"
Escalation path: Verify identity → "install this remote support tool" → credential harvest
Bank Fraud Department Pretext
"Good afternoon, this is [Name] from [Bank] Fraud Prevention. We've
detected a suspicious transaction on your account ending in [last 4
from breach data]. A purchase of $847.32 at [location]. Did you
authorize this transaction? ... No? Okay, I need to verify your
identity to block the card. Can you confirm your date of birth and
the security question on your account?"
Vendor/Partner Pretext
"Hi, I'm calling from [Known Vendor]. We're doing our quarterly
security compliance review and need to verify some access
configurations. I have your organization listed as using our
[product] — can you confirm what version you're currently running
and who manages that system on your end?"
4.4 High-Risk Targets for Vishing
| Target | Why Vulnerable | Information Sought |
|---|---|---|
| Help desk / Support | Trained to be helpful | Credentials, system info, password resets |
| Reception / Front desk | Gatekeepers with limited security training | Employee names, schedules, building layout |
| HR department | Handle sensitive employee data | PII, org structure, payroll info |
| Finance / AP | Handle payments and transfers | Banking details, approval processes |
| New employees | Unfamiliar with processes and colleagues | Everything — they don't know what's abnormal |
4.5 Notable Vishing Incidents
| Incident | Year | Technique | Impact |
|---|---|---|---|
| Twitter hack | 2020 | Attackers vished employees, convinced support staff to reset passwords | High-profile account compromise (Obama, Musk, etc.) |
| GoDaddy | 2020 | Support employees tricked into entering credentials on fraudulent login pages | Domain hijacking of cryptocurrency services |
| CEO voice deepfake | 2019 | AI-synthesized voice of German executive | $243,000 wire transfer |
| IRS scams | Ongoing | Spoofed numbers, fake badge references, threats of arrest | Millions in annual losses |
| Tech support scams | Ongoing | Apple/Microsoft impersonation | $55M+ losses in 2018 |
[CONFIRMED — SEORG Framework]
5. Smishing — SMS Phishing
5.1 Definition and Scale
SEORG defines SMiShing as "the act of using mobile phone text messages, SMS (Short Message Service), to lure victims into immediate action." Proofpoint documented a 300% increase in smishing during 2020. [CONFIRMED]
5.2 Why SMS is Effective
- 98% open rate (vs ~20% for email)
- No spam filters on most mobile devices (improving but lagging behind email)
- Shorter URLs harder to evaluate on mobile screens
- Users conditioned to act on SMS (2FA codes, delivery notifications)
- Mobile browsers hide full URLs
- Emotional association with personal device increases trust
5.3 Smishing Message Templates
Financial Alert
[BankName] ALERT: Unusual login detected on your account.
If this wasn't you, secure your account immediately:
https://bankname-secure[.]com/verify
Package Delivery
USPS: Your package #US9374728 has a delivery issue.
Update delivery preferences: https://usps-redelivery[.]com
MFA/Account Verification
Your [Company] verification code is 847291. If you did not
request this, someone may be accessing your account. Secure
it now: https://company-security[.]net
Prize/Reward
Congratulations! You've been selected for a $500 Amazon
gift card. Claim before midnight: https://amzn-rewards[.]com/claim
Government/Tax
IRS Notice: Your tax refund of $3,247.00 is pending.
Confirm your identity to process: https://irs-refund[.]com/verify
5.4 Smishing Infrastructure
- Link shorteners: bit.ly, tinyurl, custom shorteners obscure destination
- Disposable domains: Registered hours before campaign, burned after
- SMS gateways: Bulk SMS services, often abused or compromised
- Number rotation: Different source numbers to avoid carrier blocking
- Geotargeting: Messages tailored to recipient's area code/region
5.5 Impact Statistics
- FBI IC3 reported companies lost $54.2 million to social attacks including smishing in 2020 [CONFIRMED — SEORG]
- 6+ billion global smartphone subscriptions create massive attack surface
- COVID-19 pandemic massively expanded smishing (contact tracing, stimulus payments, vaccine appointments)
6. Email Phishing Taxonomy
6.1 Attack Categories
Mass Phishing
Untargeted, high-volume campaigns. Low effort per message, relies on volume.
Spear Phishing
Highly targeted using OSINT. Attackers research specific job roles and personal details via social media. Example: HR directors targeted by fake COO emails requesting payroll account changes. [CONFIRMED — SEORG]
Whaling
Targets C-suite executives with privileged access. Example: Ottawa City Treasurer received a convincing email from "her manager" requesting a $97,797 wire transfer for an acquisition. [CONFIRMED — SEORG]
Clone Phishing
Duplicates a legitimate email the target previously received, replacing attachments/links with malicious versions. Exploits existing trust in a known communication.
Business Email Compromise (BEC)
Compromises or spoofs executive email to request wire transfers, payroll changes, or sensitive data. FBI reported $2.4B in BEC losses in 2021.
Thread Hijacking
Adversaries compromise an email account, then insert malicious content into existing email conversation threads. Exploits established trust and context. [CONFIRMED — MITRE T1566]
6.2 Technical Manipulation Techniques
Domain Spoofing and Typosquatting
cornpany.cominstead ofcompany.com(m → rn)support-amazon.cominstead ofsupport.amazon.com(dash vs dot)microsoft-support.cominstead ofsupport.microsoft.com- Homograph attacks using Unicode:
аpple.com(Cyrillic 'a') - IDN homograph:
xn--pple-43d.comrenders asаpple.com
[CONFIRMED — SEORG]
Current Events Exploitation
COVID-19 generated 71,000+ reported fraud cases totaling $89.51M in losses. Healthcare saw a 580% jump in ransomware via phishing during 2020. [CONFIRMED — SEORG]
6.3 Phishing Email Anatomy
From: IT Security <security@cornpany.com> ← Spoofed/typosquat sender
Reply-To: attacker@proton.me ← Hidden reply-to
Subject: [URGENT] Password Expiration Notice ← Urgency + authority
Date: Friday 4:45 PM ← End of day/week timing
Body:
- Company logo (scraped from website) ← Visual legitimacy
- Professional formatting matching internal style ← Consistency
- Specific details (employee name, department) ← Personalization from OSINT
- Call to action with embedded link ← Payload delivery
- Deadline creating urgency ← Scarcity/fear
- Consequences for non-compliance ← Authority + threat
- Signature matching real IT staff ← Authority + legitimacy
Footer:
- "This is an automated message" ← Discourages reply
- Unsubscribe link (also malicious) ← Secondary payload
7. Physical Social Engineering
7.1 Tailgating / Piggybacking
Technique: Following an authorized person through a secured door or checkpoint without presenting credentials.
Variants:
- Hands-full approach: Carry boxes, coffee, equipment — target holds door out of politeness
- Group entry: Join a cluster of employees returning from lunch
- Distracted follow: Appear on a phone call while walking closely behind authorized person
- Delivery pretext: Carry branded packages to a door, wait for someone to open
Defense: Mantraps, badge-required turnstiles, security culture training ("it's okay to challenge")
7.2 Badge Cloning
RFID/NFC badge cloning:
- Proxmark3 or similar devices read HID/MIFARE badges at short range (1-10cm standard, extended with custom antennas)
- Badge data duplicated to blank cards in seconds
- Long-range readers concealed in bags for covert capture
- Some badges use only facility code + card number (no crypto) — trivially cloned
Countermeasures: Modern smart cards with mutual authentication (SEOS, DESFire EV3), badge + PIN, mobile credentials
7.3 Dumpster Diving
Target materials:
- Printed documents (org charts, internal memos, financial reports)
- Post-it notes with passwords
- Old hard drives and USB media
- Discarded badges and access cards
- Shipping labels revealing vendor relationships
- Building plans and network diagrams
Defense: Shredding policy (cross-cut minimum), locked dumpsters, secure media destruction, clean desk policy
7.4 Impersonation for Building Access
The 2016 Ortho Pest Control case: a hazmat suit and fake ID enabled building access and laptop theft. The Lukas Yla example (2016): researching legitimate delivery services, creating matching uniforms, and professional execution enabled infiltration of major tech companies. [CONFIRMED — SEORG]
Effective impersonation personas:
- Pest control / Exterminators (people evacuate, leaving equipment unattended)
- HVAC / Maintenance technicians
- Fire inspectors / Building code inspectors
- IT support / Cable technicians
- Delivery services (FedEx, UPS, food delivery)
Key principle: Uniforms + confidence + clipboard = access. Rarely challenged. [CONFIRMED — SEORG]
7.5 USB Drop Attacks
- Branded USB drives left in parking lots, lobbies, break rooms
- Labels that trigger curiosity: "Executive Salary Review Q4", "Layoff List 2026"
- USB Rubber Ducky / Bash Bunny for keystroke injection on plug-in
- O.MG cables that appear as normal charging cables
- Autorun payloads for data exfiltration
MITRE: T1091 (Replication Through Removable Media), T1204.002 (User Execution: Malicious File)
8. Watering Hole Attacks
8.1 MITRE ATT&CK T1189: Drive-by Compromise
Adversaries compromise legitimate websites frequented by target organizations to deliver malware. [CONFIRMED — MITRE]
8.2 Attack Chain
1. RECON — Identify websites frequented by target org/industry
2. COMPROMISE — Inject malicious code into trusted website (XSS, supply chain)
3. PROFILE — Browser/plugin fingerprinting to identify vulnerable visitors
4. EXPLOIT — Deliver browser/plugin exploit to vulnerable visitors
5. PAYLOAD — Execute code, establish persistence, begin lateral movement
8.3 Threat Group Usage
| Group | Technique | Target |
|---|---|---|
| APT28 | Custom exploit kits, reflected XSS | Government, military |
| APT32 | Compromised legitimate sites | Southeast Asian targets |
| APT37 | JavaScript profilers (RICECURRY), South Korean site compromise | South Korean organizations |
| Lazarus Group | Compromised legitimate websites | Financial sector, defense |
| Dragonfly | Custom exploit kits | Energy sector |
[CONFIRMED — MITRE T1189]
8.4 Social Engineering Element
The "social" component of watering holes:
- Target trusts the website (it's their industry conference, trade publication, vendor portal)
- No phishing email to scrutinize — the attack is invisible to the user
- Often combined with SE to drive targets to specific compromised pages ("Check out this article about our industry")
- Increasingly used with credential harvesting (fake login prompts on trusted sites)
8.5 Mitigations
- Browser sandboxing and isolation
- Exploit protection (Windows Defender Exploit Guard, EMET)
- Script blocking extensions (uBlock Origin, NoScript)
- Disable browser push notifications
- Keep browsers and plugins updated
- Network-level anomaly detection for unexpected external resource loads
[CONFIRMED — MITRE T1189]
9. Callback Phishing (BazarCall)
9.1 Concept
Callback phishing (pioneered by BazarCall/BazaCall campaigns) inverts the traditional phishing model: instead of embedding a malicious link, the email contains only a phone number. The victim initiates the call, which feels more natural and bypasses email link scanning.
Royal ransomware is documented as spreading through "phishing campaigns including callback phishing." [CONFIRMED — MITRE T1566]
9.2 Attack Flow
1. LURE EMAIL — Invoice, subscription renewal, or charge notification
"Your subscription to [Service] has been renewed for $499.99"
"Call 1-800-XXX-XXXX to cancel within 24 hours"
2. VICTIM CALLS — Target calls the number (they initiated, feels safe)
3. VISHING — Operator guides victim through "cancellation process"
"To verify your account, please visit [URL]"
"I'll need to remote into your machine to process the refund"
4. PAYLOAD — Remote access tool installed (AnyDesk, TeamViewer, ConnectWise)
OR credential harvesting via guided navigation to fake portal
OR malware downloaded under pretext of "cancellation form"
5. EXPLOITATION — Ransomware deployment, data exfiltration, lateral movement
9.3 Why Callback Phishing is Effective
- No malicious links or attachments — bypasses email security gateways
- Victim initiates contact — psychologically feels safer than inbound
- Real-time social engineering — operator adapts to target's responses
- Urgency without suspicion — financial charge creates natural motivation to call
- Legitimate remote tools — AnyDesk/TeamViewer won't trigger AV alerts
9.4 Defense
- Train users to independently verify charges (log into account directly, not via email)
- Monitor for unauthorized remote access tool installations
- Block known callback phishing phone numbers (rapidly rotating)
- Email filtering rules for common callback phishing patterns (invoice + phone number + no links)
- EDR monitoring for remote access tool deployment following user-initiated actions
10. QR Code Phishing (Quishing)
10.1 Attack Vectors
QR codes are opaque to humans — you cannot visually inspect the destination URL before scanning.
Delivery methods:
- Email: QR code embedded in email body bypasses URL scanning (it's an image, not a link)
- Physical: Fake QR codes placed over legitimate ones (parking meters, restaurant menus, conference badges)
- Documents: QR codes in printed materials, invoices, or compliance documents
- Stickers: Malicious QR stickers placed on public surfaces, ATMs, transit systems
10.2 Quishing Campaign Design
PRETEXT: "Scan to enable MFA on your account" / "Scan for Wi-Fi access"
"Updated parking payment system — scan to pay"
"Scan to access the conference materials"
QR CODE → Credential harvesting page (mobile-optimized)
→ Malware download (mobile payload)
→ OAuth consent phishing
→ Attacker-controlled Wi-Fi captive portal
10.3 Why Quishing is Growing
- QR adoption exploded post-COVID (menus, payments, check-ins)
- Most email security gateways don't decode QR codes in images
- Mobile devices have weaker security controls than desktops
- Mobile browsers show truncated URLs, hiding malicious domains
- Users are conditioned to "just scan" without questioning
10.4 SET Integration
The Social Engineer Toolkit (SET) includes a QR code generator attack vector that creates QR codes pointing to attacker-controlled infrastructure. [CONFIRMED — TrustedSec SET]
10.5 Defense
- Mobile device management (MDM) with URL filtering
- Security awareness training specifically covering QR risks
- QR scanning apps that preview URLs before opening
- Physical QR code tamper detection (does the sticker look overlaid?)
- Email security gateways with QR code image analysis capability
11. Deepfake Threats in Social Engineering
11.1 Current State of the Threat
The 2019 CEO voice deepfake case demonstrated real-world impact: AI-synthesized voice impersonating a German executive extracted $243,000 via phone call. [CONFIRMED — SEORG]
11.2 Deepfake Attack Taxonomy
Audio Deepfakes (Voice Cloning)
- Training data needed: As little as 3-5 seconds of audio (modern models)
- Sources: Earnings calls, conference talks, YouTube, voicemail greetings, social media
- Delivery: Real-time voice conversion during vishing calls
- Tools: Open-source voice cloning (RVC, Tortoise-TTS, VALL-E derivatives)
Video Deepfakes
- Real-time face swap: Used in video calls to impersonate executives
- Pre-recorded: Fake video messages from leadership
- 2024 Hong Kong case: Finance worker transferred $25M after a video call with "deepfake CFO" and multiple colleagues — all AI-generated [EXTERNAL — widely reported]
Combined Audio/Video
- Full synthetic presence in video conferences
- Fake "CEO town hall" recordings with malicious instructions
- Synthetic media for social media account impersonation
11.3 Deepfake-Enhanced SE Scenarios
| Scenario | Risk Level | Feasibility |
|---|---|---|
| Voice clone of CEO requesting wire transfer | Critical | High — proven in the wild |
| Video deepfake in Zoom/Teams call | Critical | High — demonstrated in 2024 |
| Fake voicemail from manager | High | Very High — trivial with modern tools |
| Synthetic video for identity verification (KYC bypass) | Critical | Medium — improving rapidly |
| Fake audio evidence for social manipulation | High | Very High |
11.4 Detection and Defense
Technical indicators:
- Audio artifacts: unnatural pauses, breathing patterns, background noise consistency
- Video artifacts: lip sync issues, blinking anomalies, lighting inconsistencies, edge artifacts
- Metadata analysis: generation artifacts in file metadata
Process controls:
- Out-of-band verification for any high-value request (call back on known number)
- Code word / passphrase systems for executive authorization
- Multi-party approval for financial transactions regardless of requester
- Established escalation procedures that don't rely on voice/video alone
12. Tooling and Frameworks
12.1 Social Engineer Toolkit (SET)
Repository: github.com/trustedsec/social-engineer-toolkit Author: TrustedSec (David Kennedy) Stars: 14.7k | Language: Python License: Open source penetration testing framework [CONFIRMED]
Key Modules
| Module | Purpose |
|---|---|
| Spear-Phishing Attack Vector | Craft targeted phishing emails with malicious payloads |
| Website Attack Vector | Clone websites for credential harvesting, Java applet attacks, HTA attacks |
| Infectious Media Generator | Create autorun payloads for USB/CD/DVD |
| Mass Mailer | High-volume email campaigns |
| SMS Spoofing | Send spoofed SMS messages |
| Wireless Access Point | Create rogue wireless access points |
| QR Code Generator | Generate QR codes pointing to attack infrastructure |
| PowerShell Attack Vectors | PowerShell-based payload delivery |
| Third-Party Modules | Extensible module system |
Installation: Linux native, Mac OS X, Windows WSL/WSL2 [CONFIRMED]
12.2 GoPhish
Repository: github.com/gophish/gophish Language: Go (25.5% JS for frontend) License: MIT [CONFIRMED]
Architecture
GoPhish Server
├── Campaign Management — Multi-campaign orchestration
├── Email Templates — HTML email builder with tracking
├── Landing Pages — Clone sites or build custom pages
├── User Groups — Target list management
├── Sending Profiles — SMTP configuration
├── Webhooks — Event-driven notifications
├── Reporting Dashboard — Campaign analytics and metrics
└── API — Full REST API for automation
Campaign Design Best Practices
- Baseline assessment: Run initial campaign before any training
- Realistic pretexts: Use scenarios relevant to the organization (not generic "Nigerian prince")
- Progressive difficulty: Start with obvious, increase sophistication
- Immediate training: Redirect users who click to educational content
- Metrics tracking: Click rate, credential submission rate, report rate
- No punishment: Awareness campaigns should educate, not shame
- Frequency: Quarterly minimum, monthly preferred
- Variety: Rotate vectors (email, SMS, voice) and pretexts
Deployment Options
- Binary releases (Windows, Mac, Linux)
- Docker container
- Source compilation (Go v1.10+)
12.3 CredSniper
Repository: github.com/Raikia/CredSniper (Note: repository may be archived/unavailable)
Capabilities:
- Real-time credential phishing with 2FA token capture
- Modular design supporting multiple platforms (Google, Facebook)
- API for integration with other tools
- Captures credentials AND one-time passwords in real-time before they expire
12.4 Additional Tools
| Tool | Purpose | Note |
|---|---|---|
| Evilginx2 | Reverse proxy phishing with session token capture | Bypasses MFA |
| Modlishka | Reverse proxy for real-time credential/token capture | Automated 2FA bypass |
| King Phisher | Phishing campaign toolkit | Campaign management |
| Gophish | Open-source phishing framework | Industry standard |
| Lucy | Commercial security awareness platform | Enterprise-grade |
| Cofense (PhishMe) | Commercial phishing simulation | Enterprise SAT |
13. MITRE ATT&CK Mapping
13.1 Reconnaissance (TA0043)
| Technique | ID | Description |
|---|---|---|
| Phishing for Information | T1598 | SE campaigns to extract data (not deploy malware) |
| ↳ Spearphishing Service | T1598.001 | Via third-party services (LinkedIn, etc.) |
| ↳ Spearphishing Attachment | T1598.002 | Malicious attachment to extract info |
| ↳ Spearphishing Link | T1598.003 | Malicious link to harvest credentials |
| ↳ Spearphishing Voice | T1598.004 | Voice-based phishing for information |
Mitigations: SPF/DKIM/DMARC (M1054), User Training (M1017) [CONFIRMED — MITRE]
13.2 Initial Access (TA0001)
| Technique | ID | Description |
|---|---|---|
| Phishing | T1566 | SE to gain initial system access |
| ↳ Spearphishing Attachment | T1566.001 | Malicious file attachments |
| ↳ Spearphishing Link | T1566.002 | Malicious URLs |
| ↳ Spearphishing via Service | T1566.003 | Via social media/messaging |
| ↳ Spearphishing Voice | T1566.004 | Voice phishing for access |
| Drive-by Compromise | T1189 | Watering hole attacks |
Mitigations: AV/AM (M1049), Audit (M1047), NIPS (M1031), Web Content Restriction (M1021), Anti-spoofing (M1054), User Training (M1017) [CONFIRMED — MITRE]
13.3 Execution (TA0002)
| Technique | ID | Description |
|---|---|---|
| User Execution | T1204 | Tricking users into executing malicious content |
| ↳ Malicious Link | T1204.001 | User clicks malicious URL |
| ↳ Malicious File | T1204.002 | User opens malicious file |
| ↳ Malicious Image | T1204.003 | User deploys malicious container image |
| ↳ Malicious Copy and Paste | T1204.004 | User pastes malicious code (clipboard hijacking) |
| ↳ Malicious Library | T1204.005 | User installs malicious package/library |
Key procedure examples:
- LAPSUS$: Recruited employees/contractors to provide credentials and approve MFA prompts, or install remote management software [CONFIRMED — MITRE]
- Lumma Stealer: Fake CAPTCHA instructing victims to paste Base64-encoded PowerShell into Windows Run [CONFIRMED — MITRE]
- Scattered Spider: Impersonated IT/helpdesk to instruct victims to execute remote access tools [CONFIRMED — MITRE]
13.4 Threat Actor SE Profiles
| Actor | Primary SE Vector | Notable Technique |
|---|---|---|
| APT28 (Fancy Bear) | Spearphishing, watering holes | Custom exploit kits, credential harvesting |
| Kimsuky | Spearphishing emails | Tailored emails, contact list harvesting |
| Scattered Spider | Vishing, smishing, MFA fatigue | Helpdesk impersonation, SIM swapping |
| LAPSUS$ | Insider recruitment, MFA prompt bombing | Employee/contractor bribery |
| Lazarus Group | Spearphishing, watering holes | Fake job offers, supply chain |
| APT32 (OceanLotus) | Watering holes, spearphishing | Compromised legitimate sites |
14. Defense: Security Awareness Program Design
14.1 Program Architecture
SECURITY AWARENESS PROGRAM
├── Governance
│ ├── Executive sponsorship
│ ├── Policy framework
│ ├── Budget and resources
│ └── Metrics and reporting
├── Training Delivery
│ ├── Onboarding training (mandatory)
│ ├── Annual refresher (mandatory)
│ ├── Role-specific training (targeted)
│ ├── Just-in-time training (post-failure)
│ └── Continuous micro-learning
├── Phishing Simulation
│ ├── Baseline assessment
│ ├── Regular campaigns (monthly+)
│ ├── Progressive difficulty
│ ├── Multi-vector (email, SMS, voice)
│ └── Metrics tracking
├── Culture Building
│ ├── Reporting mechanisms (easy, rewarded)
│ ├── Security champions program
│ ├── Positive reinforcement (not punishment)
│ └── Leadership modeling
└── Measurement
├── Phish click rate (trending down)
├── Report rate (trending up)
├── Time to report (trending down)
├── Training completion rate
└── Incident reduction correlation
14.2 Email Security Technical Controls
SPF (Sender Policy Framework)
- DNS TXT record specifying authorized mail servers
- Prevents simple domain spoofing
- Limitation: does not protect against display name spoofing or cousin domains
DKIM (DomainKeys Identified Mail)
- Cryptographic signature on email headers and body
- Validates message integrity and sender domain
- Limitation: does not prevent spoofing of the visible "From" field
DMARC (Domain-based Message Authentication, Reporting & Conformance)
- Policy layer on top of SPF + DKIM
- Specifies handling of failed authentication (none → quarantine → reject)
- Provides reporting on authentication failures
- Critical: Set policy to
p=rejectfor maximum protection
_dmarc.example.com TXT "v=DMARC1; p=reject; rua=mailto:dmarc@example.com; pct=100"
[CONFIRMED — MITRE T1598 Mitigations, M1054]
Additional Email Controls
- BIMI (Brand Indicators for Message Identification): Verified sender logos
- MTA-STS: Enforces TLS for mail transport
- Email gateway filtering: Sandboxing, URL rewriting, attachment analysis
- Banner warnings: External email banners, first-time sender alerts
- Link protection: URL rewriting and time-of-click analysis
14.3 Phishing Simulation Best Practices
Campaign Design Principles
- Realistic and relevant: Pretexts should match what real attackers would use against your org
- Progressive difficulty model:
- Level 1: Generic mass phishing (obvious indicators)
- Level 2: Brand impersonation with minor flaws
- Level 3: Spear phishing with OSINT personalization
- Level 4: Thread hijacking / compromised vendor simulation
- Level 5: Multi-vector (email + vishing followup)
- Immediate education: Users who fail see educational content explaining what they missed
- Positive metrics: Track and celebrate reporting rates, not just failure rates
- No gotcha culture: Shaming reduces reporting; education increases resilience
- Diverse pretexts: Rotate through authority, urgency, curiosity, fear, reward
- Timing variation: Different days, times, contexts to avoid pattern recognition
Key Metrics
| Metric | Target | Meaning |
|---|---|---|
| Click rate | <5% | Percentage who click malicious links |
| Credential submission rate | <2% | Percentage who enter credentials |
| Report rate | >70% | Percentage who report to security team |
| Time to first report | <5 min | Speed of threat identification |
| Repeat failure rate | <3% | Users who fail multiple consecutive tests |
14.4 Vishing Simulation Program
- Scope: Define which departments, what information is in-scope
- Scripts: Develop realistic vishing scripts based on threat intelligence
- Operators: Use trained operators (internal or contracted)
- Recording: Record calls (with legal approval) for training material
- Metrics: Track compliance rate, information disclosed, verification attempts
- Training: Provide immediate feedback and coaching to those who fail
- Policy: Establish and communicate verification procedures for phone requests
14.5 User Reporting Infrastructure
Requirements:
- One-click reporting button in email client (Outlook plugin, Gmail add-on)
- Clear SMS forwarding process (forward to designated short code)
- Phone reporting hotline for vishing attempts
- Positive feedback loop: "Thank you for reporting — here's what we found"
- SOC integration: reported messages feed directly into analysis queue
- Automated triage: sandbox URLs and attachments from reported messages
14.6 Role-Specific Training
| Role | Additional Training Focus |
|---|---|
| Executives | Whaling, BEC, deepfake threats, wire fraud |
| Finance/AP | Invoice fraud, payment redirection, W-2 scams |
| HR | Fake job applicants, resume malware, employee data requests |
| Help Desk | Vishing, password reset social engineering, MFA bypass |
| Developers | Supply chain attacks, typosquatting packages, fake job offers |
| Reception/Facilities | Physical SE, tailgating, impersonation, badge requests |
| Legal/Compliance | Pretexting as regulators, fake subpoenas, data disclosure requests |
14.7 Incident Response for Social Engineering
SOCIAL ENGINEERING INCIDENT RUNBOOK
TRIAGE (0-15 min)
├── Identify attack vector (phishing/vishing/smishing/physical)
├── Determine scope (who was targeted, who interacted)
├── Assess credential compromise risk
└── Classify severity (credentials entered? malware executed? access granted?)
CONTAINMENT (15-60 min)
├── Reset compromised credentials immediately
├── Revoke active sessions (all devices)
├── Block malicious URLs/domains/IPs at gateway
├── Quarantine related emails across org (clawback)
├── Disable compromised accounts if needed
└── Alert targeted users who haven't interacted yet
INVESTIGATION (1-4 hours)
├── Analyze phishing infrastructure (domain, hosting, kit)
├── Check for lateral movement from compromised accounts
├── Review email gateway logs for campaign scope
├── Search for related IOCs across environment
├── Determine if MFA was bypassed (session tokens?)
└── Timeline reconstruction
ERADICATION
├── Block all identified IOCs
├── Remove persistence (mail rules, OAuth apps, forwarding rules)
├── Patch exploited vulnerabilities
└── Update email filtering rules
RECOVERY
├── Credential reset (forced, not optional)
├── MFA re-enrollment if bypassed
├── User notification and education
└── Monitor for re-compromise (30 days)
POST-INCIDENT
├── After-action report
├── Update detection rules
├── Update phishing simulation scenarios
├── Share anonymized IOCs with industry peers
└── Adjust training based on failure patterns
Appendix A: Quick Reference — SE Attack Decision Matrix
| Scenario | Recommended Vector | Influence Principle | MITRE Technique |
|---|---|---|---|
| Credential harvest at scale | Email phishing + clone site | Urgency + Authority | T1566.002 |
| Target specific executive | Spear phishing or vishing | Authority + Scarcity | T1566.001, T1598.004 |
| Bypass MFA | Callback phishing + remote access | Helpfulness + Authority | T1566.004, T1204 |
| Gain physical access | Impersonation + tailgating | Authority + Liking | T1200 |
| Harvest info without malware | Vishing + elicitation | Reciprocity + Flattery | T1598.004 |
| Mass credential harvest | Smishing + mobile phishing page | Urgency + Fear | T1566.002, T1598.003 |
| Bypass email security | QR code phishing | Curiosity + Authority | T1566.002 |
| Target mobile users | Smishing | Urgency + Scarcity | T1566.002 |
| Long-term intelligence | Watering hole | Trust transfer | T1189 |
| High-value wire fraud | BEC + deepfake voice | Authority + Urgency | T1566, T1598.004 |
Appendix B: Red Team SE Engagement Checklist
PRE-ENGAGEMENT
[ ] Scope and rules of engagement signed
[ ] Legal review completed
[ ] Target list approved
[ ] Communication channels established with client POC
[ ] Data handling procedures agreed (credentials captured, etc.)
[ ] Emergency stop procedure defined
OSINT PHASE
[ ] Organization mapping (org chart, key personnel)
[ ] Technology stack identification
[ ] Email format discovery and verification
[ ] Social media profiling of targets
[ ] Vendor/partner identification
[ ] Physical location reconnaissance
[ ] Data breach exposure check
PRETEXT DEVELOPMENT
[ ] Persona created and documented
[ ] Infrastructure staged (domains, email, phones)
[ ] Props prepared (if physical)
[ ] Scripts developed and rehearsed
[ ] Verification resilience tested
[ ] Escalation paths planned
EXECUTION
[ ] Campaign launched per approved schedule
[ ] Real-time monitoring active
[ ] Evidence collection ongoing (screenshots, recordings)
[ ] Scope boundaries respected
[ ] Emergency stop ready
REPORTING
[ ] Timeline of all activities
[ ] Success/failure metrics
[ ] Evidence catalog
[ ] Vulnerability findings (human and technical)
[ ] Recommendations (training, process, technical controls)
[ ] Executive summary for leadership
End of module. This material is for authorized security training, assessment, and defense purposes.