Ethereum ($ETH) co-founder Vitalik Buterin (@VitalikButerin) has warned that the world is moving toward a dangerous model of artificial intelligence security, one in which a single powerful o
Ethereum ($ETH) co-founder Vitalik Buterin (@VitalikButerin) has warned that the world is moving toward a dangerous model of artificial intelligence security, one in which a single powerful organization controls the best tools.
Buterin made the remarks on November 3, 2026, in his keynote at Devcon 8 in Mumbai, titled "Open, secure and verifiable paths to a brighter future." Devcon is the flagship developer conference of the Ethereum Foundation, and this year's edition runs from November 3 to 6.
According to The Crypto Times, Buterin said slowing AI is "not happening" and warned that security through domination is becoming the default path. He described a "domination" model of AI security, in which one powerful organization holds the best tools and nobody else does.
He cautioned against concentrating powerful AI tools within a single organization. In his view, such a setup could fuel an arms race as excluded groups seek access to those tools. Without open and verifiable defensive technology, he said, the world will drift toward a model where one powerful group holds the best tools and everyone else races to catch up.
Buterin did not rule out all forms of restraint. He said there is a role for "pacing the frontier," which he described as not advancing frontier AI models so quickly that defensive work cannot keep up.
As an alternative, Buterin presented defensive, decentralized acceleration, known as DAC. It is built on open-source software, verifiable hardware, and cryptographic proofs.He described DAC as an effort to "aggressively improve our security" across physical, digital, information, and biological domains.
The talk also covered the technical building blocks. He called for open-source software, formally verified security properties, fewer trust dependencies, and cryptographic proofs that let users prove specific facts about themselves without disclosing everything else. He applied the same logic to hardware, including devices crypto users rely on to store private keys. "We need more open and more verifiable silicon," he said.
The concept is not new. Buterin laid out the philosophy in a 2024 essay on techno-optimism and revisited it a year later. In that follow-up, he defined the core idea of d/acc as decentralized and democratic differential defensive acceleration. At the time, he said his caution centered on superintelligent AI and the risk of doom or irreversible human disempowerment if the technology is built in the wrong ways.
At Devcon 8, he returned to the same theme with a sharper warning: without a better approach, domination could become the default.
Sources
