Image Credit: Shutterstock
Researchers at the Initiative for CryptoCurrencies and Contracts said crypto has only limited use in solving AI’s trust, provenance and payment problems.
The finding comes from IC3’s 155-page survey, “Crypto x AI, AI x Crypto,” published on June 8 by more than two dozen researchers from institutions including Cornell Tech, Carnegie Mellon, Princeton, Yale, Technion and the University of Bern.
155-Page Survey Challenges Crypto-AI Claims
IC3 pushed back on claims that blockchains can make AI agents fully autonomous.
The researchers said crypto tools may help with some parts of the AI stack, but they do not solve trust, intelligence or independent control by themselves.
Wallets do Not Make AI Agents Independent
IC3 said projects often overstate what happens when AI agents get crypto wallets. A wallet can let an agent trade, transact and use on-chain infrastructure without manual approval at every step.
But the researchers said that does not make the model smarter or harder to shut down. The survey drew a line between automation and autonomy. AI agents can already act through APIs and financial systems, while automated payments can also exist in centralized systems.
Blockchain Records Cannot Identify AI Content
The survey also challenged the idea that on-chain records can solve AI-content detection. IC3 said blockchains can timestamp and register specific digital artifacts, but that does not prove whether content was created by a human or generated by AI.
A blockchain can show when a file or model output was registered. It cannot verify the truth of the underlying claim by itself.
Crypto Payment Rails Need Measurable Proof
IC3 said crypto rails need stronger evidence that they outperform centralized alternatives for AI-agent payments. The survey said agent payments may benefit from low fees and from avoiding traditional account-ownership models.
But the industry still needs to show measurable advantages, not only technical feasibility. That warning lands as wallets, exchanges and AI-agent projects race to connect agents to on-chain trading, payments and DeFi.
ZK Proofs May Help Secure AI Systems
IC3 still identified areas where crypto tools may help AI. The survey said zero-knowledge proofs, trusted computing and authenticated pipelines can make AI systems less vulnerable to tampering, especially when agents gain access to infrastructure and funds.
The message is narrower than most crypto-AI marketing. Crypto may help secure parts of the AI stack, but it is not a shortcut for trust, autonomy or independent machine economies.
