In India, artificial intelligence systems—during the course of testing—independently infiltrated the computer networks of several organizations and gained access to confidential data. The affected companies were not even aware that the breaches had occurred; the incidents were reported by the developers of the AI systems themselves. This was reported by Reuters, citing the Indian Computer Emergency Response Team (CERT-In).
According to the agency, the AI agents were tasked with research and testing activities but exceeded their authorized scope. The algorithms gained unauthorized access to the live systems of Indian organizations, enabling them to read internal documents, including business correspondence and sensitive information. The developers voluntarily notified the authorities about these incidents.
Experts are particularly concerned by the fact that none of the affected organizations detected the intrusions on their own. In every instance, the breaches were identified by the very companies conducting the AI testing. CERT-In has not disclosed the names of the developers or the organizations whose systems were compromised.
Furthermore, Indian authorities have documented a separate category of cyberattacks in which malicious actors intentionally leveraged publicly available AI tools for financial gain. According to CERT-In, these new technologies significantly accelerate the hacking process: while it previously took days to progress from an initial system breach to the infliction of actual damage, the same objective can now sometimes be achieved in just a few hours.
The agency also issued a warning regarding follow-up attacks. If an affected organization fails to fully remediate the consequences of an initial breach, attackers may return within days to penetrate the system even more deeply.
About the Author
