Follow us on Google News
NEW DELHI — India’s cybersecurity agency, the Indian Computer Emergency Response Team, has been granted access to an unreleased artificial intelligence model developed by United States-based firm Anthropic to strengthen national cyber defence. Known as Mythos, the advanced model is being deployed to detect security flaws across critical government information technology systems and digital infrastructure, while also recommending technical patches to fix them. Government sources indicate that the system possesses sophisticated capabilities to uncover technical vulnerabilities in software networks before malicious actors and cybercriminals can exploit them to stage attacks.
The strategic deployment reflects an expanding role for artificial intelligence tools within the country’s core cybersecurity framework. The agency is relying on the tool not merely for scanning digital perimeter risks, but to formulate targeted solutions for vulnerabilities that could otherwise allow unauthorised access to sensitive official databases. While the initiative offers higher technical efficiency compared to traditional human-led assessments, the operational structure of the tool has prompted internal evaluation over how national digital assets are audited.
Cloud Delivery Prompts Security Concerns
Despite the performance advantages offered by the advanced model, its deployment has raised specific data protection concerns among authorities. Because Mythos is accessed through an external cloud-based platform, government officials remain cautious about exposing technical details concerning critical public systems to third-party digital environments. The fundamental sensitivity of vulnerability data means that any unauthorised exposure of infrastructural weaknesses could inadvertently offer hostile entities the precise blueprints needed to compromise state systems.
Consequently, managing digital boundaries and maintaining rigorous data confidentiality have emerged as primary considerations during the model’s active deployment. Officials are evaluating technical safeguards to prevent sensitive infrastructure telemetry and configuration data from remaining vulnerable during diagnostic scans. Ensuring that diagnostic inputs do not compromise overall system privacy remains central to the broader rollout of the tool across federal administrative departments.
Deployment of Alternative AI Models
To mitigate potential operational risks associated with relying on a single external service, the national cybersecurity agency has simultaneously engaged multiple other artificial intelligence platforms. Government sources observe that these supplementary models deliver approximately 80 to 85 per cent of the technical capabilities provided by Mythos. By distributing the analytical workload among various platforms, the agency seeks to reduce structural dependence on one foreign cloud framework while limiting exposure risks.
The diverse deployment also allows security teams to compare the analytical rigor, accuracy, and data-handling protocols of different automated tools before applying their findings to vital networks. While alternative models may present variations in functionality, having multiple avenues provides essential redundancy. This multi-model approach enables investigators to address potential system gaps systematically, balancing advanced automated discovery against necessary confidentiality requirements across critical installations.
Project Glasswing and Preventive Defence
Anthropic has restricted access to Mythos strictly to select global organisations and public agencies through an initiative known as Project Glasswing. The primary mandate of the programme centres on identifying software weaknesses across high-priority IT networks and resolving them proactively. Through this structured access, participating bodies are provided technical recommendations to neutralise complex flaws prior to weaponisation by external cyber threat groups.
The adoption of such capabilities arrives at a moment when cyber intrusions are growing increasingly sophisticated worldwide. While traditional vulnerability detection relies extensively on specialized personnel analyzing manual processes, advanced artificial intelligence can process vast technical frameworks with greater speed and breadth. For the Indian cybersecurity establishment, the central challenge going forward involves leveraging these advanced discovery capabilities while maintaining strict sovereignty and control over sensitive information relating to core digital assets.
