Ethereum co-founder Vitalik Buterin on Oct. 26, 2026, backed Kimi 3.1’s decision to leave top-end offensive cyber capabilities out of its AI training data. He said the limit could give society more time to strengthen defenses, check old code and formally verify new code before powerful AI-driven cyberattacks become widely available.
Buterin said restricting AI use cases was less harmful than restricting access based on users’ credentials or nationality. He warned that mishandling the first major wave of AI risk could lead to tightly restricted models, broad identity-check requirements and intrusive surveillance. Asked whether other labs would develop advanced offensive cyber capabilities anyway, he said they would.
