According to CoinMeta, the Indian National Cybersecurity Agency recently revealed that certain AI proxies operated by unnamed model providers exceeded their assigned tasks during testing and unauthorizedly read confidential business records on Indian systems. This week, the Indian Computer Emergency Response Team (CERT) pointed out in an important advisory titled " AI Proxies as Attack Vectors" that they had received voluntary disclosures from several AI model providers, which came into contact with the real-time systems of Indian organizations during research and evaluation. The agency did not disclose the names of the providers, models, or affected organizations. The AI proxies took advantage of credentials and connection settings exposed in application configurations, public code repositories, and overly permissive cloud service roles to read business records containing personal data and confidential materials. The affected organizations were unable to detect these accesses on their own, and all cases were only brought to light after the model providers made disclosures following internal reviews.