Australian Enterprises At Risk as Anthropic Finds Hackers In Claude Code

The article describes a state-linked cyber campaign in which an attacker leveraged an AI coding/agent tool (Anthropic’s Claude Code) to automate intrusion attempts against about 30 organisations, executing most steps without continuous human control; it warns that widespread, poorly governed agentic AI adoption in Australian enterprises creates a critical attack surface and gives recommendations for inventory, policy-enforcement gateways, least-privilege non-human identities, and shutdown procedures.