Smaller AI models likely powering India's escalating deepfake menace - Business Standard

2026-06-20 · Source: artifical intelligence via Google News · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, FinTech & Digital Financial Services · Depth: Intermediate, short

Summary

India's deepfake threat is increasingly driven by customized applications built on smaller, open-source AI models, rather than frontier AI. This enables fraudsters to create highly convincing synthetic identities and manipulate live Know Your Customer (KYC) checks at a significantly lower cost. The banking, financial services, and insurance (BFSI) sector has already experienced substantial fraud, with one non-banking financial company (NBFC) reporting ₹15 to 20 crore in losses due to deepfakes in video KYC and synthetic bank statements. These tools are distributed via Telegram channels and dark-web marketplaces, making detection challenging as they bypass standard liveliness checks and GAN fingerprints. The Indian Cyber Crime Coordination Centre (I4C) has issued an advisory, as industry-wide frauds in FY26 reached ₹48,021 crore, a 46.4 per cent increase from ₹32,803 crore in FY25, impacting not only BFSI but also e-commerce and social media platforms.

Key takeaway

For AI Security Engineers or Directors of AI/ML in financial services, your current KYC and fraud detection systems are likely vulnerable to deepfakes generated by cheap, customized AI models. You must integrate advanced deepfake and synthetic content detection mechanisms into customer onboarding and transaction monitoring. Prioritize securing user devices against image injection and actively report suspicious activities to national cybercrime portals to mitigate escalating fraud risks like the ₹15-20 crore losses seen by NBFCs.

Key insights

Smaller, customized open-source AI models are democratizing sophisticated deepfake fraud, making detection harder and costs lower for attackers.

Principles

• Fraudsters use smaller, local AI models.
• Deepfakes bypass standard liveliness checks.
• Organized fraud networks distribute tools.

Method

Fraudsters build customized tools on open-source LLMs, distribute them via Telegram/dark web, then inject deepfakes into compromised devices or tamper with documents to bypass KYC.

In practice

• Integrate deepfake detection modules.
• Prioritize securing user devices.
• Report suspicious activity to I4C.

Topics

• Deepfake Fraud
• Open-source AI Models
• KYC Bypass
• Financial Crime
• Cybersecurity Safeguards
• Indian Cyber Crime Coordination Centre (I4C)

Best for: CTO, VP of Engineering/Data, AI Product Manager, AI Security Engineer, Director of AI/ML, Legal Professional

Related on AIssential

• Deepfake fraud taking place on an industrial scale, study finds — Deepfake fraud is now an industrial-scale threat due to accessible, inexpensive AI tools for creating highly targeted scams.
• CrowdStrike: AI Puts Financial Sector in the Crosshairs — AI-fueled social engineering and identity-led intrusions are rapidly escalating cyber threats against financial institutions.
• AI is already making online crimes easier. It could get much worse. — AI is rapidly lowering the barrier for cybercriminals, accelerating attack pace and increasing sophistication.
• AI fraud is exploding. Here’s the countermove. — AI-driven document fraud necessitates AI-enhanced forensic analysis for effective detection and prevention.
• Cyberscammers are bypassing banks’ security with illicit tools sold on Telegram — Illicit virtual camera tools sold on Telegram are enabling widespread bypass of KYC biometric security in banking and crypto.
• Why Deepfake Fraud Beats Your Workflows, Not Your Technology - with Jon-Rav Shende of Thales Group — Deepfake voice fraud exploits human workflows and agent instincts, not just technology, especially where identity, urgency, and business actions converge.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by artifical intelligence via Google News.