Ethereum co-founder Vitalik Buterin has warned that AI poses significant risks to blockchain security and privacy, even as it offers tools to improve user experience and system design.
Buterin Warns AI Could Be ‘Security and Privacy Nightmare’ for Ethereum
1 April 2026 - 18:20 CEST
By Matt Neill
His remarks come as Ethereum developers accelerate work on privacy-focused infrastructure and explore how AI can be integrated into onchain applications.
"AI is by default a security and privacy nightmare," Buterin said via video at EthCC 2026 in Cannes, France, during a discussion focused on Ethereum privacy initiatives, including Kohaku, a privacy-first tooling for the Ethereum ecosystem.
Risks and opportunities for Ethereum
Buterin said AI introduces new attack surfaces and privacy challenges, particularly when sensitive data interacts with external models or centralized systems.
At the same time, he said that properly designed AI systems could improve the execution of private transactions and make advanced cryptographic tools more accessible to users.
"AI done well can actively and greatly improve the user experience of private execution," he said.
He added that Ethereum developers should prioritize building expertise in combining privacy, security and usability, as these will become increasingly important across both blockchain and broader digital systems.
"I hope we continue to look at AI carefully and take this as an opportunity to achieve both security and privacy at a much higher level than before, for both advanced and new users," he said.
Push for open, secure AI infrastructure
Buterin also called for a decentralized approach to AI development that aligns with Ethereum’s core principles of censorship resistance and user control.
He said the ecosystem should focus on building open, secure and privacy-preserving AI systems rather than relying on centralized providers.
"There’s a need for not just an open, decentralized AI space, but a censorship-resistant, open source, private and secure AI space," he said.
He suggested that developers should prioritize local AI models where possible, reducing reliance on external systems that may introduce additional risks.
Where remote models are required, he said multiple layers of safeguards should be used to ensure safe interaction and minimize exposure.
The remarks highlight a growing intersection between AI and blockchain development, as both technologies evolve and increasingly overlap.
Buterin’s comments suggest Ethereum’s next phase may depend not only on scaling and adoption, but also on how effectively it integrates emerging technologies like AI while maintaining its core focus on decentralization, privacy and security.
