Romance Scams Using AI Chatbots Surge to $10 Billion

Americans lost $1.14 billion to romance scams in 2023, a figure that has since ballooned as organized crime syndicates shift from human-operated fraud to automated AI chatbots [1]. These large language models (LLMs) now manage thousands of fake romantic relationships simultaneously, significantly increasing the success rate of "pig butchering" crypto schemes [1].

The shift to automated fraud

The transition from human-led scams to AI-driven operations has fundamentally altered the economics of digital fraud. Previously, criminal networks in regions like Myanmar and Cambodia relied on trafficked workers to manually manage conversations across social platforms [1]. Today, these syndicates utilize LLMs to maintain 24-hour, emotionally adaptive personas that remember personal details and mirror a victim's tone without fatigue or error [1].

Research confirms that this automation is not only cheaper but more effective at extracting funds. A study published in 2026 found that AI agents achieved a 46% compliance rate when making financial requests, compared to just 18% for human operators [1]. By removing the need for a large, human workforce, criminal groups have reduced the marginal cost of each new "relationship" to near zero, allowing for massive scaling of their operations [1].

Industrialized pig butchering

The "pig butchering" model—or sha zhu pan—functions by building long-term emotional intimacy before pivoting to a fraudulent crypto investment platform or a direct financial request [1]. While human-run operations were once limited by language barriers and shift changes, AI-powered centers now operate continuously in multiple languages [1].

The scale of this infrastructure is expanding rapidly. The UNODC reported a 600% increase in deepfake-related criminal services advertised on platforms like Telegram between February and July 2024 [1]. Furthermore, the U.S.-China Economic and Security Review Commission noted that the integration of AI image generation and LLMs has allowed these syndicates to create more convincing personas that are increasingly difficult for victims to distinguish from real individuals [1].

What to watch

• Compliance Metrics: Monitor whether the gap between AI and human success rates in financial solicitation continues to widen as models become more contextually aware [1].
• Regulatory Enforcement: Watch for further actions against AI providers, similar to the February 2025 move by OpenAI to ban accounts originating in Cambodia that were generating fraudulent romance scam content [1].
• Platform Security: Observe if dating and social apps implement new verification protocols to counter the rise of non-human accounts that can sustain months-long conversations [1].

As these criminal syndicates continue to upgrade their toolsets, the distinction between genuine human interaction and automated deception is becoming increasingly blurred. The shift from manual labor to industrial-scale AI automation suggests that romance-based financial fraud will remain a high-volume threat for the foreseeable future [1].