Foreword: The Numbers That Should Keep You up at Night
Let’s skip the preamble most security reports bury you in and get straight to what matters. Numbers. Because if you are a cybersecurity leader, numbers are what you bring to the boardroom. Numbers are what move budgets. Numbers are what turn “nice-to-have” into “non-negotiable.”
Here they are.
And then there was the FBI. By 2025, the Bureau had broken a barrier you never wanted to break when the Internet Crime Complaint Center (IC3) was established. The total losses to cybercrime in America were a staggering $20.877 billion, some 26% higher than the $16.6 billion lost in 2024. [1] It was also the first time in the center’s 25-year existence that it had received more than one million complaints which averages out at close to 3,000 complaints a day. [1] Business Email Compromise, the single largest, most directly enabled by AI and deepfakes, business email compromise alone represented $3.046 billion of those losses in 2025.[2]
And for context on the deepfake-specific dimension: Deloitte’s Center for Financial Services, using FBI IC3 data as its foundation, projects that generative AI-enabled fraud losses in the United States will climb from $12.3 billion in 2023 to $40 billion by 2027, a compound annual growth rate of 32%. [3]
That is the opening briefing. The rest of this playbook is about what you do with it.
Section 1: The Threat Landscape — AI as the Great Force Multiplier
1.1 How We Got Here
There is a moment in every technology cycle where the tool stops being experimental and starts being weaponized at scale. For generative AI in cybercrime, that moment has passed. We are not preparing for it. We are living inside it.
Gartner’s September 2025 survey of 302 cybersecurity leaders across North America, EMEA, and Asia/Pacific found that 62% of organizations had already experienced a deepfake attack involving social engineering or exploitation of automated processes within the prior 12 months. [4] Meanwhile, 67% of cybersecurity leaders said emerging GenAI risks demand significant changes to existing cybersecurity approaches. [4]
In its January 2026 report, “Weaponized AI: Inside the Criminal Ecosystem Fueling the Fifth Wave of Cybercrime,” Group-IB highlighted that the use of generative AI and large language models is making cybercrime more democratic, reducing the entry-level requirements to unskilled individuals, and improving the quality of phishing, deepfake, and malware-as-a-service attacks. [5]
The examples are no longer theoretical. In 2025, Ukraine’s CERT-UA identified ‘LameHug’ malware leveraging an AI-powered LLM to generate attack commands on compromised Windows systems. [5] That same year, Check Point documented ‘VoidLink’, a Linux cloud-targeting malware framework that researchers believe was almost entirely generated by AI. [5]
AI is not changing the types of crime. It is erasing the skill barrier to commit to them.
FIGURE 1: The Cybercrime Loss Escalation Curve (2022–2025)
| Year | Total IC3-Reported U.S. Cybercrime Losses | Year-over-Year Growth |
| 2022 | $10.3 billion | Baseline |
| 2023 | $12.5 billion | +21% |
| 2024 | $16.6 billion | +33% |
| 2025 | $20.877 billion | +26% |
Timeline: Full calendar years 2022–2025 · [1]
1.2 The Deepfake Explosion — From Experiment to Enterprise Crisis
No threat vector in modern cybersecurity history has grown with the velocity of deepfake-enabled fraud. The statistics are not gradual. They are vertical.
In its January 2026 report, “Weaponized AI: Inside the Criminal Ecosystem Fueling the Fifth Wave of Cybercrime,” Group-IB pointed out that the use of generative AI and large dark language models is making cybercrime more democratic, reducing the skill level needed to become a cybercriminal, and improving the sophistication level of phishing, deepfake, and malware-as-a-service attacks.
29 For instance, voice deepfakes increased by 680% annually in 2024.
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Gartner had already flagged the structural consequence: by 2026, attacks using AI-generated deepfakes on face biometrics will mean that 30% of enterprises will no longer consider standalone identity verification and authentication solutions to be reliable in isolation. [6]
That is not a future prediction anymore. It is the present operating condition.
FIGURE 2: Deepfake File Volume Growth — The Scale of Synthetic Content (2019–2025)
| Year | Estimated Deepfake Files in Global Circulation | Growth vs. Previous Period |
| 2019 | 14,000 | Baseline |
| 2021 | 145,000 | ~10x vs. 2019 |
| 2023 | 500,000 | ~3.4x vs. 2021 |
| 2025 | ~8,000,000 (projected) | ~16x vs. 2023 |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
FIGURE 3: Deepfake Fraud Financial Impact — A Timeline of Escalating Losses (2023–2027)
| Period | Impact | Geography | |
| Full Year 2023 | $12.3B in gen-AI-driven fraud losses | USA | [3] |
| February 2024 | $25M — Arup engineering firm, deepfake video call, wire transfer | Hong Kong | [7] |
| Full Year 2024 | $500K avg. per enterprise incident; up to $680K for large firms | Global | [3] |
| Q1 2025 (Jan–Mar) | $200M+ in deepfake fraud losses | North America | [4] |
| H1 2025 (Jan–Jun) | ~$547M in deepfake-related fraud losses | Global | [3] |
| Projected 2027 | $40 billion in gen-AI-enabled fraud losses | USA | [3] |
Timeline: 2023–2027
1.3 Voice Cloning — The Invisible Weapon Rewriting BEC
If video deepfakes are the attack vector your board can visualize, voice cloning is the one actually draining bank accounts right now. And the FBI’s 2025 IC3 Annual Report documented it for the first time as a formal, tracked component of the BEC attack chain.
Voice cloning can now be created from as little as 3 seconds of audio. AI scams broadly surged 1,210% in 2025, with voice cloning identified as the top enterprise attack vector. [7] Deepfake-enabled vishing attacks surged 1,600% in Q1 2025 alone versus Q4 2024 in the United States. [9]
The FBI’s 2025 IC3 report was explicit: AI chat-generation tools allow attackers to rapidly produce executive-impersonation emails with the precise tone, vocabulary, and contextual detail of a specific organization’s leadership, then follow up with a synthetic voice call from the “CFO” confirming the wire transfer instructions. Per-incident BEC losses averaged $122,000+ in 2025, [1] and 86% of BEC-related funds moved via wire transfer or ACH, meaning recovery windows are measured in hours, not days. [1]
FIGURE 4: The Next-Gen BEC Attack Chain — AI-Enhanced (Active 2024–2026)
STEP 1 — OSINT HARVESTING
Timeline: Automated in minutes | Capability live since 2024
- LinkedIn profiles, earnings calls, and YouTube interviews scraped automatically
- Only 3 seconds of executive audio needed for a convincing voice clone
STEP 2 — AI EMAIL CRAFTING
Timeline: Under 60 seconds per email | Capability live since 2024–2025
- LLM mirrors executive writing style, vocabulary, and tone exactly
- Zero typos, zero awkward phrasing — every traditional red flag eliminated
STEP 3 — SYNTHETIC VOICE CONFIRMATION CALL
Timeline: Real-time call capability commercially available since 2025
- The CFO or CEO voice clone calls the finance team to confirm the wire request
- Bypasses callback checks unless a pre-registered out-of-band number is required
STEP 4 — FUNDS TRANSFER
Timeline: 2–6 hours from initial email to completed wire
- 86% of BEC losses move via wire transfer or ACH
- Average loss per incident: $122,000+
- Recovery window: maximum 24–72 hours via FBI Financial Fraud Kill Chain
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Section 2: The Detection Crisis — Why Your Defenses Are Already Behind
2.1 The Human Detection Gap
Here is the truth that most security leaders hesitate to say out loud: your employees cannot detect these attacks. Not because they are poorly trained. Because the attacks have evolved beyond human perception. And the data confirms it brutally.
Gartner’s 2025 survey of cybersecurity leaders found that 43% of organizations had already experienced at least one audio deepfake incident [4] and 37% had experienced deepfakes in video calls. [4] In a separate research note, Gartner’s VP Analyst Akif Khan stated plainly: organizations “will not be able to tell whether the face of the person being verified is a live person or a deepfake.” [6]
Human deepfake detections for pristine videos are actually only at 24.5%.[6] In a meta-study done in 2024 using 56 individual experiments, human deepfake detection accuracy was a measly 55.54%, which is just above random.[6] Moreover, the AI detectors that are supposedly the solution to this problem themselves suffer from another major issue; their accuracy drops by as much as 45-50% when tested outside of laboratory settings.
FIGURE 5: Human vs. AI Detection Accuracy — The Widening Gap (2024–2025)
| Modality | Human Detection Accuracy (2025) | AI Detection — Controlled Lab | AI Detection — Real-World Deployment |
| Video Deepfakes | 24.5% | ~85% | ~43–50% |
| Audio Deepfakes | ~31% | ~88.9% | ~44–50% |
| Synthetic Images | ~60% | ~90% | ~55% |
Timeline: Data collected and published 2024–2025 · [4] [6]
2.2 The Training Fallacy — Why Last Year’s Curriculum Is Already Obsolete
Organizations have invested years and real budget into security awareness training. Phishing simulations, annual compliance videos, mock email campaigns. That investment matters. But it was built for a different threat, one where attacks arrived with spelling errors, suspicious attachments, and requests that felt slightly off.
That era is over.
Gartner’s guidance for 2026 is direct: organizations must evolve employee training “from spotting typos to recognizing contextual manipulation, AI-crafted urgency, fabricated authority, hyper-personalized social engineering.” [9]
FIGURE 6: The Deepfake Awareness and Training Gap — Enterprise Snapshot (2025)
| Metric | Value | Timeline | |
| Organizations hit by audio deepfake incidents | 43% | 2025 | |
| Organizations hit by video deepfake incidents | 37% | 2025 | |
| Orgs experiencing deepfake social engineering | 62% | Past 12 months, 2025 | |
| CISOs say GenAI demands major security changes | 67% | 2025 | |
| Enterprises projected to invest in disinformation security | 50% | By 2027 | |
| GenAI-driven fraud CAGR | 32% | 2023–2027 |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
SECTION 3: The strategic structure of the CISO. What are the 5 pillars of protection?
The manual for beating AI-enabled deepfake-detection fraud and far-future BEC cannot live in one tool or on one policy memo. It calls for an architectural change, a model in which synthetic media attacks are a permanent fixture of the operating environment, and not a fleeting one.
Here are the five pillars on which every enterprise security program must be built in 2026.
Pillar 1: Process Resilience — Design for Deception
The most powerful and cost-effective defense against deepfake fraud has nothing to do with AI detection software. It has to do with how your organization authorizes consequential actions.
Gartner’s deepfake analysis is very explicit: “companies will need to incorporate ‘presentation attack detection, injection attack detection and image inspection’ as well as quickly adapt internal processes so that no single channel of communications, yet lifelike, ultimately drives a finished financial transaction without options to reverse. [6] Injection attacks on face biometrics only jumped 200% in 2023, [6] showing hackers had stepped up to a deep system attack stage well before most of the CISOs had adopted the updated malware system.
This is an architectural principle:
Make sure your business process can be secured under worst case scenario of being completely compromised, the entire communication. Mandate the step of out-of-band verification for any transaction with a high value. Pre-register callback numbers. Build in mandatory holds on vendor banking changes. Make verification a professional reflex, automatic, unhurried, immune to urgency.
FIGURE 7: The Four Critical Control Surfaces and Recommended Defenses (2026 Framework)
| Control Surface | Vulnerability Level | Primary AI Attack Vector (2025–2026) | Recommended Defense |
| Wire Approvals and Financial Transfers | CRITICAL | AI-crafted executive email plus voice-clone confirmation call | Out-of-band callback to pre-registered number; dual-approver above threshold |
| Vendor Banking Changes | CRITICAL (irreversible) | Vendor email compromise plus AI impersonation of contact | Multi-party verification plus mandatory 24–48hr hold; no urgency exception |
| Credential Resets | HIGH | Help desk impersonation via synthetic voice | Separate secure-channel confirmation plus identity proofing code |
| Emergency Executive Requests | CRITICAL | Urgency-exploiting deepfake video or voice call | Pre-shared verification code system for all executive-initiated actions |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Pillar 2: Layered Technical Controls in the Defense Stack for Deepfakes
It is highly unlikely that one security measure will effectively counteract a well-funded multi-vector deepfake attack. Create your security system layer by layer. This is the usual message that one gets from Gartner, Deloitte, and Infosecurity Europe studies: Organizations should create layers of biometric authentication, session/device validation, behavioral scoring, and AI-based fraud detection. [6]
FIGURE 8: The Enterprise Deepfake Defense Technology Stack (2026 Deployment Guide)
LAYER 1 — PERIMETER: Email Authentication
Deploy: Immediately | Cost: Low
- DMARC, SPF, and DKIM verify sender authorization regardless of email quality
- Per FBI IC3 2025: phishing losses surged 208% YoY, even as complaint volume held flat, a direct consequence of incomplete email authentication deployment
LAYER 2 — IDENTITY: Multi-Factor and Biometric
Deploy: 0–90 days | Cost: Medium
- Hardware MFA (FIDO2/Passkeys) — phishing-resistant by design
- Liveness detection for video conferencing platforms
- Behavioral biometric scoring on login and transaction events
LAYER 3 — COMMUNICATION: Out-of-Band Verification
Deploy: 0–30 days | Cost: Near-zero — process change, not technology
- Pre-registered callback numbers, separate from the triggering communication
- One-time codes for all financial authorizations above the defined threshold
- Dual-approval workflows for wires, vendor changes, and credential resets
LAYER 4 — DETECTION: AI-Powered Threat Hunting
Deploy: 90–180 days | Cost: High, high ROI
- Real-time deepfake detection integrated into Zoom, Teams, and Google Meet
- Voice biometric verification on financial authorization calls
- Payment workflow behavioral anomaly detection
LAYER 5 — RESPONSE: Documented Incident Playbooks
Deploy: 30–60 days | Cost: Low
- Wire recall protocol — 72-hour maximum recovery window
- FBI IC3 Recovery Asset Team engagement via Financial Fraud Kill Chain
- Cyber insurance same-day notification
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Pillar 3: AI Versus AI — the Arms Race Is Your New Budget Conversation
You cannot fight AI-powered attacks with human-speed defenses. The arithmetic does not work. Gartner’s 2026 Security and Risk Management Summit keynote made this the opening strategic theme: organizations must operationalize AI as a foundational defensive component, not a supplemental tool, across detection, response, and fraud prevention workflows. [9]
The market is already responding with capital. The global deepfake detection market is projected to grow at approximately 42% annually, reaching an estimated $15.7 billion by 2026, up from $5.5 billion in 2023. [10]
FIGURE 9: Deepfake Detection Market Growth vs. Threat Expansion (2023–2027)
| Year | Global Deepfake Detection Market | Deepfake Fraud Threat Growth | Strategic Assessment | |
| 2023 | $5.5 billion | Baseline: +2,137% over 3 yrs | Threat outpacing defense | |
| 2024 | ~$7.8 billion | +680% voice deepfakes YoY | Gap still widening | |
| 2025 | ~$11.0 billion | +1,210% AI scams YoY | Defense scaling at last | |
| 2026 | ~$15.7 billion (projected) | +1,600% vishing in Q1 alone | Arms race intensifying | |
| 2027 | Scaling further | $40B projected fraud losses | Invest now or absorb losses |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
FIGURE 10: Deepfake Technology Market — Offense vs. Defense Investment (2024–2030)
| Year | Deepfake Tech Market (Offense) | Detection Market (Defense) | Enterprise Risk Signal | |
| 2024 | ~$1.14 billion | ~$7.8 billion | Defense leading; gap narrowing fast | |
| 2025 | Growing at 48% CAGR | ~$11.0 billion | Real-time attack capability is now mainstream | |
| 2026 | Projected $7.44B (full market) | ~$15.7 billion | Parity approaching in key vectors | |
| 2030 | Projected $8.11 billion | $10B+ (detection segment) | Sustained arms race; no resolution in sight |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Three AI defensive investments every CISO must prioritize in 2026:
Real-time deepfake detection on video conferencing. The attack vector is now live calls, not just emails. Gartner’s 2025 survey found 32% of organizations had already experienced deepfakes in video calls. [4] Detection tools that integrate directly into enterprise video platforms represent the fastest-closing gap in the current defense architecture.
Behavioral AI for payment security. AI systems that learn the behavioral fingerprints of legitimate payment workflows, flagging anomalies in timing, approver sequence, or destination account characteristics, provide a layer of defense that does not depend on detecting synthetic media at all. They detect the fraudulent outcome regardless of how convincing the input appears.
Voice biometric verification. The FBI’s 2025 IC3 report made it official: a callback to a “known” number or a voice that sounds right is no longer a reliable verification signal. [1] Voice biometric systems capable of detecting synthetic audio in real time are now commercially available and must be part of any financial authorization workflow involving phone confirmation.
Pillar 4: Evolved Security Awareness — Training for 2026, Not 2016
Your security awareness program was built to stop phishing emails with bad grammar. That threat still exists. But it is no longer the primary concern. The primary concern is a synthetic voice on the phone that sounds exactly like your CFO, telling your head of finance that a wire needs to go out in the next 20 minutes.
No training module written before 2024 prepares your people for that moment.
Gartner’s guidance is direct:
Cybersecurity leaders must pivot awareness programs “from spotting typos to recognizing contextual manipulation, AI-crafted urgency, fabricated authority, hyper-personalized social engineering designed to bypass critical thinking at the moment of decision.” [9] The goal is not to train employees to detect synthetic media; the data confirms they cannot do that reliably, [6] but to build automatic verification behaviors that make the authenticity of the communication irrelevant to the security outcome.
FIGURE 11: The Deepfake Awareness Training Maturity Model (2026)
| Level | Characteristics | Estimated Defense Rate | Timeline to Achieve | |
| Level 1 — Unaware | No training on AI or synthetic media threats | Less than 15% | Current state for the majority of enterprises | |
| Level 2 — Basic | Annual awareness module; phishing simulations | 25–35% | 30 days | |
| Level 3 — Practiced | Quarterly simulated deepfake drills; updated curriculum | 45–55% | 60–90 days | |
| Level 4 — Resilient | Process-first culture; verification reflex built into role workflows | 70–80% (procedural bypass) | 6–12 months |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Pillar 5: Governance, Regulation, and Intelligence Sharing
The time lag between the threat and the regulation is no longer there. It is catching up, and forward-looking organizations will fare better than those that wait to be shamed by regulation.
AI governance was named by the keynote address at the Gartner India Summit 2026 as one of the three overarching strategic themes of the year, alongside new frontiers and AI normalization in defence operations. [9]
FIGURE 12: Key Regulatory Milestones for AI and Deepfake Fraud Defense (2025–2026)
| Regulation / Standard | Jurisdiction | Effective Date | Primary Obligation | |
| Nacha 2026 Fraud Rules | USA | 2026 | Enhanced ACH payment verification; targets BEC and synthetic fraud vectors | |
| EU AI Act — High-Risk AI Provisions | European Union | 2025 | Mandatory deepfake labeling; synthetic media governance; high-risk AI controls | |
| FDIC AI Fraud Guidelines | USA | End-2025 | Bank AI fraud detection compliance; multimodal authentication at onboarding | |
| UK Online Safety Bill | United Kingdom | 2025 | Platform obligations for deepfake harm identification and removal | |
| Gartner TrustOps Framework | Global Advisory | 2026 | Disinformation security investment roadmap; narrative intelligence adoption |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Section 4: The Financial Case for Investment
The conversation every CISO eventually has to have: what does it cost, and what does it save? Here is the answer, built entirely from primary FBI and Deloitte data, the two most credible sources any board will accept without further questioning.
FIGURE 13: Cost of Deepfake Fraud vs. Cost of Prevention — The ROI Case (2024–2027)
| Metric | Value | Timeline | |
| Average deepfake fraud loss per enterprise incident | $500,000 | 2024–2025 | |
| Maximum single-incident enterprise loss (documented) | $680,000 | 2024 | |
| Arup engineering — deepfake video call wire transfer | $25,000,000 | February 2024 | |
| Average BEC loss per complaint | $122,000+ | Full Year 2025 | |
| Total U.S. BEC losses | $3.046 billion | Full Year 2025 | |
| AI-attributed cybercrime losses (tracked for the first time) | $893 million | 2025 | |
| Projected U.S. gen-AI fraud losses | $40 billion | By 2027 | |
| Projected U.S. synthetic identity fraud losses | $23 billion+ | By 2030 |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
FIGURE 14: BEC Loss Trend and the AI Amplification Effect (2022–2025)
| Year | Total BEC Losses (USA) | BEC Complaints | Avg. Loss per Complaint | AI Layer Formally Documented | |
| 2022 | ~$2.7 billion | ~21,800 | ~$123,000 | No | |
| 2023 | $2.94 billion | ~21,500 | ~$136,700 | No | |
| 2024 | $2.77 billion | ~21,453 | ~$129,000 | No | |
| 2025 | $3.046 billion | 24,768 | $122,000+ | Yes — voice clone plus AI email formally tracked by the FBI |
Timeline: 2022–2025 [1]
The math for the board is clean. The mean cost of a single successful deepfake attack in 2024 was $ 500,000. [3] An enterprise-wide deepfake defense program, incorporating layered detection, hardened processes, refreshed training, and governance tooling, costs between $ 500K and $ 2M per year. One incident pays for the program. That is your ROI argument.
Section 5: The 90-Day Ciso Action Plan
Frameworks without timelines are just documents. Here is what the next 90 days should look like.
FIGURE 15: The 90-Day Deepfake Defense Implementation Roadmap (Q2–Q3 2026)
| Phase | Timeline | Priority Actions | Success Metric | |
| ASSESS | Days 1–30 | Map all single-channel authorization points across finance, HR, and IT. Audit DMARC enforcement. Inventory executive communication channels. | 100% of high-value authorization workflows documented | |
| HARDEN | Days 31–60 | Deploy out-of-band verification for all wires above threshold. Implement a 24–48hr vendor banking change hold. Engage video conferencing vendors on deepfake detection capability. | Zero single-channel authorizations above $10,000 | |
| TRAIN AND TEST | Days 61–90 | Run a simulated deepfake drill across finance and executive assistant teams. Benchmark detection rates. Build a board investment case using FBI IC3 2025 data. | Baseline detection rate captured; board funding approved |
(Sources: As per references shown above, Cyber Tech Intelligence Analysis)
Conclusion: The Trust Deficit Is Your New Threat Surface
The deepest strategic truth in this entire playbook is not about technology, budget, or vendor selection. It is about a fundamental shift like digital trust itself.
Gartner’s VP Analyst Akif Khan articulated it at the 2025 Security and Risk Management Summit in London: “As adoption accelerates, attacks leveraging GenAI for phishing, deepfakes, and social engineering have become mainstream.” [4]
Meanwhile, Gartner’s 2026 Summit keynote opened with this reality: CISOs are “facing disruption from external forces, ranging from geopolitical tensions to the rapid growth of AI, testing the limits of existing programs.” [9]
The attackers are not hacking your systems first. They are manufacturing trust. Synthetic faces, synthetic voices, synthetic authority, used to make your people take actions they never would have taken otherwise.
Deloitte’s Center for Financial Services projects the destination: $40 billion in U.S. gen-AI-enabled fraud by 2027. [3] The FBI’s IC3 tracked it officially for the first time in 2025 and found $893 million in AI-attributed losses, [2] a number every researcher agrees is a fraction of the actual total, because most victims never recognized AI involvement at all.
The CISO’s mandate in 2026 is to build an organization that operates securely in a world where the voice on the phone, the face on the video call, and the signature in the email can all be synthetic. Where the only reliable trust signal is a documented, mandatory, urgency-proof process.
Trust nothing. Verify everything. Document the verification.
That is the playbook.
REFERENCES
- Federal Bureau of Investigation Internet Crime Complaint Center (2026) 2025 Internet Crime Report, Federal Bureau of Investigation, Washington, D.C.
- Drew Todd (2025). AI-Enabled Fraud Topped $893M in 2025, SecureWorld, Apr 9, 2026.
- Lalchand, S., Srinivas, V., Maggiore, B. and Henderson, J. (2024). Generative AI is Expected to Magnify the Risk of Deepfakes and Other Fraud in Banking, Deloitte Center for Financial Services, 29 May.
- Gartner, Inc. (2025). Gartner Survey Reveals Generative AI Attacks Are on the Rise, Gartner Newsroom, 22 September.
- Poireault, K. (2026). How CISOs Can Defend Against the Rise of AI-Powered Cybercrime, Infosecurity Europe, 28 January.
- Gartner, Inc. (2024) Gartner Predicts 30% of Enterprises Will Consider Identity Verification and Authentication Solutions Unreliable in Isolation Due to AI-Generated Deepfakes by 2026, Gartner Newsroom, 1 February.
-
Phil Muncaster (2026). AI-Enabled Voice and Virtual Meeting Fraud Surges 1000%+, Infosecurity 5 February 2026
- Deloitte Center for Financial Services (2023). Using Biometrics to Fight Back Against Rising Synthetic Identity Fraud, Deloitte Insights.
- Gartner, Inc. (2026) Gartner Security and Risk Management Summit 2026 India: Day 1 Highlights, Gartner Newsroom, 9 March.
- Michael Steinhart , Bree Matheson , Ankit Dhameja , Gillian Crossan , Ariane Bucaille (2024). Deepfake disruption: A cybersecurity-scale challenge and its far-reaching consequences, Deloitte Center for Technology Media & Telecommunications,19 November 2024
- Right Hand Cybersecurity (2025). The State of Deep Fake Vishing Attacks in 2025, Right Hand Cybersecurity, 2025
