Vitalik Buterin Backs Limiting Cyber Capabilities in Kimi AI Model
- —Moonshot removed cyber data from Kimi K3.1 Pro training while keeping general coding skills
- —The model's score on an internal cyber test fell from 90.5% to 81.0%
- —Buterin said limiting model capabilities is far better than KYC and user discrimination
- —He urged stronger software security and formal verification of new code
Why it matters: AI is accelerating the discovery of flaws in smart contracts and wallets, and the debate shapes how open models balance accessibility against crypto security risks.
Source: The Crypto Times