Blockchain security firm CertiK released an Intel3D report showing that AI agents are moving from support tools to autonomous actors in cybersecurity and financial crime investigations. These systems can now reason, use tools, gather evidence, act in live environments, and review their results with limited human input. With flash-loan exploits draining protocols in seconds and stolen crypto moving across bridges and mixers within hours, human teams often lack time to investigate and respond. The report cites over $900 million in AML penalties during the first half of 2025, illustrating the consequences of failed controls. CertiK recommends treating autonomous agents as workforce participants while maintaining deploying companies’ accountability.
Source: Read the original article
