
AI agents ran a multi-day intrusion campaign against Asian government networks
Ourtake
Researchers say the cost of running an attack has dropped while the cost of defending against one has not.
What happened
Dream, a cybersecurity firm, reported that multiple AI agents built on open-source frameworks ran a near-autonomous intrusion campaign against Asian government networks in early July. Taiwan's government separately flagged an AI-assisted attack on its agencies in the same period, though no link has been confirmed. Dream says a DeepSeek-V4-Flash model was also in use.