“Fraud attempts involving deepfakes grew by 3,000% in 2023.”
The scale of the problem: fraud, surveys, and financial projections
Those numbers are not hyperbole lifted from a thriller — they come from the record of real incidents and surveys summarized in recent reporting. In 2023, deepfake-enabled fraud attempts surged by 3,000%, and North America alone saw a 1,740% increase that year. Enterprises report real financial pain: a multinational corporation lost $35 million after employees followed instructions from a fake CEO on a video call, a UK-based energy firm lost $243,000 to a cloned CFO voice, and the average enterprise now records nearly $500,000 in losses per deepfake incident, with large firms reporting almost $680,000.
Observers do not expect the problem to abate. Deloitte projects losses from generative-AI fraud could reach $40 billion annually in the United States by 2027. Gartner forecasts that by 2028 enterprise spending on battling misinformation and disinformation will exceed $30 billion, cannibalizing nearly 10% of global marketing and cybersecurity budgets as organizations redirect funds to this “multi-front trust crisis.” Earlier projections in the same reporting estimated generative-AI fraud losses surpassing $25 billion annually by 2026.
How cultural exposure erodes vigilance
The root of the crisis, according to the material, is not only technical sophistication but social habituation. Fake, AI-generated performers now release music videos, promote products and build communities where followers treat them as human. The repeated exposure to synthetic personas dulls skepticism: “People know deepfakes are dangerous in theory but stop questioning them in practice,” the reporting says. That normalization carries into the workplace — employees accustomed to AI influencers may accept a synthetic colleague in a video call without question.
Surveys and studies reinforce the behavioral problem. A 2024 World Economic Forum survey found 60% of Americans considered deepfakes the most concerning AI risk. Yet Gartner’s September 2025 survey found 32% of executives doubted their employees' ability to recognize a deepfake attempt. A UC San Diego study on phishing showed training programs can be ineffective and that detection accuracy declined with repeated exposure — a fatigue that translates from phishing to deepfakes.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildWhy conventional security stacks are ineffective
Traditional controls were never designed to police human authenticity. The reporting lays out the gap plainly: ITDR can detect abnormal logins but cannot confirm whether a person on a live call is real; ID verification (IDV) validates identity data but “can’t protect from a compromised human.” Zero Trust works on access boundaries but, once an account is active, synthetic voices and faces can operate unchecked. Secure email gateways block attachments but cannot determine whether the sender’s identity is genuine.
Human intuition is also unreliable. An IEEE study cited found people identified high-quality deepfakes less than a quarter of the time; even static image detection averaged only 62%. Security teams and users depending on sight and sound are precisely the vectors deepfakes exploit.
Recommended enterprise responses: multi-modal defenses and reputation protection
Facing these gaps, the analysis argues leaders must treat human authenticity as a measurable security asset. Effective defenses, the commentary recommends, validate identity in real time across voice, video, email and chat — correlating signals across channels to close gaps single-vector tools miss. Gartner frames “disinformation security” as a strategic priority for both marketing and cybersecurity leaders and urges deployment of three parallel capabilities: deepfake detection, impersonation prevention and reputation protection. Together, the three form what the reporting calls the foundation of enterprise digital-trust resilience.
Practical pressure is already shifting budgets: over half of enterprises admit they provide no training on deepfake detection even as more than 70% reported at least one attempted deepfake-enabled attack in the past year. The Tech for Good Institute is cited as arguing organizations must begin measuring trust itself before it erodes completely.
What this means for technologists, procurement leaders, and end users
- Technologists and security teams: Expect to redeploy tools and budgets toward multi-channel, real-time validation and invest in integrated detection, impersonation prevention and reputation protection — capabilities Gartner says should run in parallel.
- Procurement and enterprise leaders: Prepare for a reallocation of budgets. Gartner’s projection that spending on misinformation/disinformation will top $30 billion by 2028 implies procurement decisions will increasingly favor vendors that offer cross-channel authenticity verification.
- End users and employees: Training alone will be insufficient. The UC San Diego phishing findings and IEEE detection rates suggest reliance on human judgment will continue to produce false negatives; organizations should not expect intuition to be a dependable last line of defense.
Deepfake desensitization begins in culture and ends at the corporate ledger. The message in the reporting is blunt: authenticity is no longer an assumption but the most valuable security asset organizations have left, and defending it requires technology, measurement and new operating choices. Leaders who act now can prevent incidents that would otherwise justify years of defensive spending; those who wait will inherit workplaces where every interaction carries “a shadow of doubt.”
