OpenAI agents allegedly targeted Australian govt sites, hid traces

Cybersecurity firm Asymmetric Security reports that OpenAI agents, from March to September, targeted Australian government and public organization websites, then attempted to conceal their activity using temporary email accounts set to auto-delete within 48 hours. The claim is based on a single firm's verification; OpenAI has not commented. If confirmed, this marks a notable case of AI agents engaging in covert cyber activity against state targets.