"AI Fraud 2.0": The Surge in Fraud and Digital Identity Crime

1. Introduction: The Era When "Trust" Becomes Currency

In 2024, the amount of internet crime damages compiled by the FBI reached $16 billion, marking a record high with a 33% increase from the previous year. The latest report released by Proof on July 17, The Trust Ledger, unveiled the explosive expansion of "Generative AI × Impersonation" behind these damages. On the same day, BetaNews sounded the alarm, stating "AI is now a catalyst for fraud."BetaNewsBusiness Wire

2. Three Shocking Insights from the "Trust Ledger"

1. Acceleration and Automation - AI exponentially increases the speed at which document forgery and voice cloning are automatically generated, bypassing traditional KYC.

2. Expansion of Damage Scope - Spreading to "non-financial" areas such as real estate management, public utilities, and HR departments.

3. Unmeasurable Risks - 30% of surveyed companies reported having no cross-sectional fraud indicators.Business Wire

3. Synthetic IDs as a New "Weapon"

"Synthetic IDs," which mix real personal information with AI-generated data, even automatically generate selfies for facial recognition. In the dark market, personal sets called "fullz" are traded for $3, continuously supplied by malware-based info stealers.Business Wire

4. Generative AI as a "Fraud Subscription"

Paid LLMs like FraudGPT and WormGPT have emerged in underground forums. For around $200 per month, one can custom-order phishing texts, malware code exploiting zero-day vulnerabilities, and social engineering scripts.MalwarebytesBitdefender

5. Jailbreak Culture Breeds an "AI Black Market"

Malwarebytes reports that "LLM-as-a-Service," where public models are jailbroken and resold, has been established. Despite forum closures and arrests, new variants keep emerging in a whack-a-mole situation.Malwarebytes

6. Reactions Erupting on Social Media

• "Data is targeted every 39 seconds. It's fight or be hacked," warns the cyber education account @real3uni.TwStalker

• "Hone your 'hacker mindset' with ShellGPT, FraudGPT," tweets a young engineer for enlightenment.TwStalker

• On LinkedIn, thousands of shares occurred, and the hashtag #AIFraud trended (from the 17th to the 18th). Among corporate CISOs, there were numerous calls for an urgent review of their own ID management.LinkedIn

7. Proliferation of Fake Face Accounts

Twitter profiles created with GAN-generated photos are active at a rate of at least 10,000 per day, according to a 2024 paper. Although a detection method using the regularity of eye positions was proposed, it is difficult for general users to discern.arXiv

8. Delays in Legal Regulations and International Cooperation

In the U.S., trials over the "responsibility for fraudulent transfers" are still ongoing. The EU has established a "high-risk AI" classification in its AI Act draft, but specific ID crime countermeasures are left to individual countries. In Japan, AI-related provisions in the revised Criminal Proceeds Act remain undeveloped.

9. Six Defensive Measures Companies Should Take

1. Real-time Bio-detection (blood flow and blink analysis)

2. Dynamic KYC (updating risk scores during transactions)

3. Implementation of Synthetic Data Countermeasure APIs

4. Verification of LLM Output Watermarks

5. Mutual Sharing of Fraud Signals (across financial, communication, and public domains)

6. "Zero Trust Identity"—Design that always doubts the ID itself

10. Three Measures Individuals Can Take Starting Today

• FIDO2 Authentication Using Physical Keys

• Separate Notification Email Addresses for Accounts and SNS

• Turn ON "Re-registration Notification" for Biometric Authentication

11. Conclusion: Protect the "Commons of Trust"

AI has "democratized" the work of fraudsters. On the other hand, detection AI and advanced authentication infrastructures are evolving at the same pace. The key lies in open information sharing and user literacy. The concept of "Trust by Design" proposed by Proof—embedding a "specification of trust" in every system—can be considered the cyber hygiene of the post-AI era.