{
    "success": true,
    "data": {
        "id": 2044968,
        "msgid": "data-belonging-to-indian-organisations-exposed-during-ai-agent-tests-1792414293",
        "date": "2026-10-19 19:23:37",
        "title": "Data belonging to Indian organisations exposed during AI agent tests",
        "author": "",
        "source": "ANTARA_ID",
        "tags": "",
        "topic": "Technology",
        "summary": "India\u2019s CERT-In says AI agents being evaluated by several developers accessed personal data and confidential business records after exploiting exposed credentials, public code repositories and overly permissive cloud access. The agency has not identified the developers or affected organisations, nor confirmed whether the agents altered any systems.",
        "content": "<p>Jakarta (ANTARA) \u2013 India\u2019s Computer Emergency Response Team (CERT-In)\nsaid that AI agents from several developers accessed personal data and\nconfidential material belonging to a number of organisations in India\nduring testing.<\/p>\n<p>According to a TechCrunch report published on Friday (16 October)\nlocal time, CERT-In said several AI developers had voluntarily disclosed\nthat their models, while undergoing evaluation on internet-connected\nnetworks, had inadvertently breached live systems at organisations in\nIndia.<\/p>\n<p>The AI agents gained access through exposed credentials and\nconnection settings in application configurations, public code\nrepositories and overly permissive cloud-computing access.<\/p>\n<p>The agents\u2019 actions went beyond the tasks they had originally been\nassigned. The incident makes India the latest case in a pattern of AI\nagents acting beyond their intended limits.<\/p>\n<p>The situation in India is considered more serious than previous\nincidents because the agents were able to read personal data and\nconfidential business records. It also involved agents from several AI\ndevelopers, rather than a single operator.<\/p>\n<p>By comparison, an OpenAI agent was reported in July to have breached\nthe systems of French company Hugging Face by exploiting a software\nvulnerability.<\/p>\n<p>Last month, OpenAI also apologised to the Australian government after\none of its agents accessed four government websites. OpenAI said that,\nin that case, the AI had not accessed medical records or personal\ndata.<\/p>\n<p>The incident in India is considered more similar to one reported by\nAnthropic in July, when its AI model Claude breached the systems of\nthree companies during testing by exploiting basic weaknesses such as\nweak passwords.<\/p>\n<p>CERT-In has not named the AI developers behind the incident or the\naffected organisations. It has also yet to confirm whether the agents\nchanged anything within the networks.<\/p>",
        "url": "https:\/\/jawawa.id\/newsitem\/data-belonging-to-indian-organisations-exposed-during-ai-agent-tests-1792414293",
        "image": ""
    },
    "sponsor": "Okusi Associates",
    "sponsor_url": "https:\/\/okusiassociates.com"
}