The potential losses that autonomous AI agents could cause in crypto may end up dwarfing anything the sector has seen, industry leaders warn. An AI agent, the relevant term here, is software that acts independently: it executes trades, signs transactions, and interacts with protocols without a human approving each step. Those same leaders say today's billion-dollar hacks could end up looking like pennies.
Crypto leaders have identified three specific hurdles blocking widespread agentic adoption: trust, hallucinations, and unclear legal liability.
Trust sits first on that list because the agentic model requires granting authority in advance. A user or protocol deploying an agent authorizes it to act before any specific decision is reviewed. That preemptive handoff of signing power is what makes agents useful. It is also what opens a new attack surface: the model itself. The billion-dollar exploits that have marked crypto's recent history required attackers to find a specific protocol flaw. An agent given equivalent signing authority makes the AI the target. Industry leaders say the scale of potential losses in an agentic world could leave those existing figures behind.
What hallucinations mean for an industry where transactions are final
A hallucination, in AI terminology, is when a system produces a wrong answer delivered with the same confidence as a correct one. For a text-generating model, that produces a bad paragraph. For an agent authorized to sign transactions and move assets on-chain, it produces a bad trade or a misdirected transfer, with the agent as the only decision-maker. Blockchain transactions generally cannot be reversed, which turns the accuracy gap into direct capital risk.
The liability question that law has not caught up with
Legal liability is the third hurdle named by crypto leaders. When an autonomous agent executes a bad trade or sends assets to the wrong address, who absorbs the loss? Responsibility could fall on the developer who built the agent, the protocol it touched, or the user who deployed it, and current law does not resolve the question cleanly. That ambiguity, alongside hallucinations and trust, is what crypto leaders say has to be solved before agentic adoption can proceed.