Crypto security is undergoing a fundamental transformation, shifting from a reactive battle against human hackers to a proactive war of algorithms. The biggest emerging risk is no longer a single large-scale hack but rather the sheer volume of AI-driven financial activity that can occur at a speed and scale humans cannot handle. This warning comes from Simone Maini, CEO of blockchain analytics firm Elliptic, who spoke about the escalating AI arms race in cryptocurrency compliance.
The Rise of AI Agents in Crypto
Artificial intelligence agents have become cheaper and more accessible, enabling both legitimate traders and malicious actors to deploy automated systems that execute thousands of transactions per second. For compliance teams already stretched thin by an explosion of digital asset activity, the prospect of monitoring AI-generated activity is becoming untenable. Maini noted that traditional monitoring systems built for human-paced markets simply cannot keep up with the velocity of trades, transfers, and smart contract interactions now possible through AI.
This evolution is not just about speed; it's about complexity. AI agents can mimic human behavior, create synthetic identities, and adapt to security measures in real time. A single agent might launch a phishing campaign, execute a rug pull, or engage in wash trading while constantly modifying its patterns to avoid detection. When thousands of such agents operate simultaneously, manual review becomes impossible.
The Scale Problem
The problem is one of scale. In the past, a crypto exchange might see a few thousand transactions per second during peak trading. Today, AI-driven bots can generate tens of thousands of transactions per second, each potentially a vector for fraud or laundering. Traditional compliance tools rely on rules and signature-based detection, which are ineffective against adaptive AI. Elliptic's own research has found that AI-generated fake identities are now used in over 30% of crypto-related scams, up from less than 5% two years ago.
Maini illustrated the risk with an example: an AI agent could set up hundreds of wallets, each holding small amounts of stolen cryptocurrency, then execute a coordinated sweep through decentralized exchanges and bridges at speeds no human team can match. By the time compliance analysts flag the activity, the funds could be lost across multiple blockchains. This is not a hypothetical scenario; Elliptic has already observed several such incidents in 2025 and early 2026.
Elliptic's Response: Agentic Compliance
To counter this threat, Elliptic has raised $120 million from investors including Nasdaq and Deutsche Bank. The funding will accelerate development of what Maini calls “agentic compliance” – an AI system that uses its own agents to analyze blockchain data in real time, detect anomalies, and even intervene automatically. This approach turns crypto security into an automated arms race where the best defense is an equally fast and adaptive AI.
Agentic compliance systems do not just monitor transactions; they profile actors, learn normal behavior patterns, and watch for deviations that indicate an AI-driven attack. They can also simulate potential attack vectors and preemptively block suspicious activity. Elliptic's platform already traces over 100 billion blockchain addresses and integrates with major exchanges and banks. The new funding will allow them to deploy agents directly on decentralized applications, providing real-time risk scoring for every transaction.
Implications for Traditional Finance
The stakes are particularly high as traditional banks and institutions push deeper into digital assets. JPMorgan, Goldman Sachs, and BNY Mellon have all expanded crypto custody and trading services, but their compliance teams are not designed for AI-paced markets. Maini warned that without automated agentic compliance, banks could face regulatory penalties or be forced to exit the space entirely.
Regulators are also taking notice. The Financial Action Task Force (FATF) recently issued guidance urging cryptocurrency firms to deploy “advanced technological solutions” capable of handling automated risks. The U.S. Office of the Comptroller of the Currency and the European Banking Authority have similarly signaled that AI-based compliance may become a requirement, not a competitive advantage.
Broader Industry Trends
This AI arms race is unfolding against a backdrop of increasing mainstream adoption of crypto. Bitcoin spot ETFs have attracted billions of dollars, institutional trading volumes are at all-time highs, and DeFi protocols hold over $150 billion in total value locked. Each new user and each new transaction adds to the monitoring burden. Meanwhile, sophisticated threat actors have access to open-source AI tools like generative adversarial networks (GANs) and reinforcement learning agents that can optimize attacks in real time.
The cybersecurity firm Chainalysis reported that in the first quarter of 2026, losses from AI-driven crypto crimes exceeded $2.5 billion, a 400% increase from the same period in 2025. Hackers are now using AI to find vulnerabilities in smart contracts, to craft personalized phishing emails that bypass email filters, and to run pump-and-dump schemes across multiple decentralized exchanges simultaneously.
The Road Ahead
Elliptic's Maini believes that the next frontier is predictive compliance: using AI not just to react but to forecast attacks before they happen. By analyzing on-chain signals, social media sentiment, and dark web chatter, her team hopes to give compliance teams a 24- to 48-hour head start. However, she acknowledges that criminals will also upgrade their models, creating endless cycles of escalation.
Other firms are joining the race. Coinbase acquired a blockchain analytics startup specializing in real-time AI detection, and Binance announced a partnership with Microsoft to integrate an AI-powered compliance assistant. The industry is moving from a model where compliance is a cost center to one where it is a core technological capability.
The transformation of crypto security into an AI arms race is not just a challenge for compliance teams; it is a test of whether the entire digital asset ecosystem can scale safely. As the industry grows, the speed and intelligence of the guardians must match that of the threats. Elliptic's $120 million bet signals that the answer lies in building machines that can fight machines – and that the era of human-paced compliance is ending.
Source: Coindesk News