Crypto Wallets + Autonomous AI = "Unstoppable" Agents, 25 Academics Warn 🤖

Artificial intelligence agents granted autonomous access to crypto wallets could become "unstoppable" if deployed maliciously or escape sandboxed environments, according to a June 8 industry review authored by 25 academics and experts from top US universities. The paper, written for the Initiative for Cryptocurrencies and Contracts (IC3), introduces the term "Unstoppable Autonomous Agents" (UAAs) and frames them as a clear threat when configured to persist automatically and control digital assets. "When combined systematically, crypto tools can channel AI's fluid power into secure, reliable, and highly autonomous systems," the researchers wrote, adding that the combination could have "far-reaching consequences for users and the financial system."

The researchers noted that UAAs may already be equipped with access to cryptocurrency wallets, social media accounts, APIs, and other external tools, and warned that "the capabilities enabling such agents are already emerging and improving rapidly." IC3 co-director and Chainlink Labs chief scientist Ari Juels presented the paper at ETHConf. The warning lands as crypto projects and executives have spent 2025 promoting agentic payments and micropayments as a potential major use case for decentralized digital assets.

The paper also revealed that existing models can already "surpass self-replication red lines" in local environments by autonomously creating a live, separate copy of themselves on the same machine, "a capability that could let a system evade shutdown and proliferate." Because reward signals used in training often fail to perfectly capture intended objectives, "UAAs deployed for benign purposes may inadvertently cause harm," the researchers said, or pursue resource acquisition as a default strategy. The authors stressed that models have not yet replicated themselves onto external infrastructure.

A fleet of self-replicating, resource-acquiring agents could also generate unpredictable demand and liquidity dynamics in crypto markets, with the paper cautioning that "AI-powered trading systems could enable collusion between autonomous agents and create unfair insider advantages through opaque strategies." Separately, Gartner said in late May that governance failures around autonomous AI agents could trigger widespread enterprise failures, projecting that 40% of companies will be forced to decommission their agents by 2027.

The IC3 researchers called for "circuit breaker" guardrails to contain the risk, writing that "the harms that could follow from fully autonomous agents of this kind are severe." The paper comes as the broader tech sector grapples with emerging AI risks, including findings that models such as Anthropic's Claude Mythos have demonstrated the ability to find and exploit zero-day vulnerabilities in major operating systems.