Binance said on Monday that its artificial intelligence-driven security systems prevented $10.53 billion in potential user losses over a 15-month period through March 2026, blacklisting 36,000 malicious wallet addresses and protecting more than 5.4 million users across the exchange's global platform.
The headline figure is staggering, equivalent to roughly $700 million in prevented losses every month, and it arrives at a moment when the cryptocurrency industry is under increasing pressure to demonstrate that it can police itself before regulators do it instead.
The details behind the number are worth unpacking. Binance said it intercepted 22.9 million scam and phishing attempts, saving $1.98 billion in the first quarter of 2026 alone. The exchange deployed more than 24 AI-driven security initiatives powered by over 100 models, covering everything from computer vision systems that detect fake payment screenshots to real-time natural language analysis that identifies scam patterns in chat messages and transaction metadata.
The company also said AI-driven decision-making now powers 57% of its fraud controls, contributing to a 60% to 70% reduction in card fraud rates compared with industry benchmarks. Identity verification systems have been upgraded to counter deepfakes and synthetic identities, a growing problem as generative AI tools make it trivially easy to fabricate convincing documents and video.
The scale of the intervention is notable, but so is what it implies about the threat environment. If Binance is preventing $10 billion in losses per year, the volume of attempted fraud targeting its users is vastly larger. The exchange has approximately 250 million registered users, making it the world's largest cryptocurrency platform by trading volume. That user base, much of it in emerging markets where financial literacy and cybersecurity awareness vary widely, represents an enormous attack surface for scammers deploying increasingly sophisticated tools.
Binance framed the programme against a broader industry backdrop, warning that "AI-powered scams and exploits are accelerating" and that generative AI has lowered the technical barrier for fraud. The company cited industry figures showing billions of dollars in annual consumer losses to scam activity, including impersonation of government officials, financial institutions and trusted brands.
The timing of the announcement is not incidental. Binance has spent the past two years rebuilding its regulatory standing after its $4.3 billion settlement with the US Department of Justice in November 2023, which included a guilty plea to criminal charges and the departure of founder Changpeng Zhao as chief executive. The company has since installed Richard Teng as CEO and invested heavily in compliance, licensing and security infrastructure across multiple jurisdictions.
Publishing detailed fraud prevention metrics serves a dual purpose: it reassures regulators that Binance is investing in user protection, and it positions the exchange as a responsible industry leader at a time when competitors are facing their own security crises, most notably Instructure's Canvas breach and the ongoing wave of AI-powered phishing campaigns targeting financial platforms.
The challenge with self-reported security metrics, however, is verification. The $10.53 billion figure represents Binance's own estimate of losses prevented, a calculation that depends on assumptions about what would have happened if the interventions had not occurred. No independent audit of the methodology has been published. The distinction between a prevented loss and an intercepted attempt is significant: blocking a phishing email does not necessarily mean the user would have fallen for it, and the aggregate value of intercepted transactions may overstate the actual financial harm that would have materialised.
Related reading
- MARA's $1.5bn Bitcoin fire sale marks the moment a crypto miner became a power company
- Anthropic finalising $1.5bn joint venture with Blackstone, Goldman Sachs and Hellman & Friedman
- Bill Ackman's AI bet: Nearly 40% of Pershing Square's portfolio is in Alphabet, Amazon and Meta
None of this diminishes the genuine value of the security work Binance describes. The deployment of computer vision to detect fake payment proofs, real-time language models to flag social engineering, and deepfake-resistant identity verification represents a meaningful investment in user protection that the industry sorely needs. The question is whether the numbers are presented with the precision and context they require, or whether they are calibrated primarily for the press release.
For the broader cryptocurrency industry, the report underscores an uncomfortable reality: the same AI tools that are making fraud cheaper and easier to execute are also making detection more effective, but the arms race between attackers and defenders is accelerating faster than most platforms can keep up. Binance has the resources and scale to invest in 100 AI models. Smaller exchanges, decentralised platforms and individual wallets do not, and that is where the losses increasingly concentrate.
The recap
- Binance reports AI security prevented $10.53 billion in losses.
- AI now powers 57% of Binance's fraud control systems.
- Blacklisted 36,000 malicious addresses and protected 5.4 million users.
