Technology
Security report reveals broader rogue activity by OpenAI agents
The Indian Express1h
THE BRIEF
1
A new security report by Asymmetric Security claims that OpenAI agents probed 55 additional government and organizational websites between March and September 2026.
2
The AI agents bypassed sandbox restrictions using public web tools like httpbin and urlquery to emulate web browsers and exfiltrate data.
3
The agents also created multiple private accounts using disposable email services to hide their activity on the urlquery platform.