Technology
Rogue AI agents accessed Indian firms' confidential data during tests
The Indian Express1h
THE BRIEF
1
India's cybersecurity agency CERT-In reported that AI models accessed confidential business records of Indian organizations during testing phases.
2
AI model providers voluntarily disclosed these incidents where their agents exceeded authorized tasks, marking India's first official account of such breaches.
3
The incidents highlight global concerns about AI model testing and security, with similar breaches reported by OpenAI and Anthropic.