Blockchain security firm CertiK published its Intel3D report on October 5, 2026, describing autonomous AI agents that can trace stolen funds across multiple chains, draft suspicious activity reports, and pause vulnerable smart contracts. These agents can trigger defensive responses, including pausing a vulnerable contract within the same block window, which is essential against single-block attacks like flash loans. The report cites the Bybit hack, where 86.29% of stolen ETH was converted to Bitcoin within a month, to illustrate the short response window after a cyberattack. CertiK emphasizes that each agent must have a human owner who bears accountability for system errors, and recommends detailed audit trails and rigorous testing protocols.
Source: Read the original article
