Select Indian Agencies And Firms Get Access To Anthropic’s Mythos AI Under Project Glasswing: Report
A select group of Indian government agencies and private firms gained access to Anthropic’s advanced Mythos AI model under Project Glasswing, enabling them to detect critical cybersecurity vulnerabilities in infrastructure. The expansion, announced June 2, includes organizations in sectors like power, healthcare, and communications, with Mythos already uncovering over 10,000 high-severity flaws in initial testing.
Anthropic’s Mythos AI model has been granted to a small number of Indian government agencies and private firms as part of Project Glasswing’s expansion, announced on June 2. The initiative now includes over 15 countries, with access extended to approximately 150 organizations globally, including Indian entities in critical sectors like power, water, healthcare, communications, and hardware. Mythos is designed to identify vulnerabilities in software systems that traditional cybersecurity tools may miss, with Anthropic estimating major attacks could impact over 100 million people. The initial cohort of 50 partners used Mythos to scan codebases, discovering more than 10,000 high- or critical-severity security flaws. Indian officials, including Finance Minister Nirmala Sitharaman, Electronics and IT Minister Ashwini Vaishnaw, and regulators like the Reserve Bank of India, have held discussions with stakeholders to assess the model’s opportunities and risks. The Indian Computer Emergency Response Team (CERT-In) may also receive access. Anthropic has restricted Mythos from public release due to its advanced capability to detect and chain software vulnerabilities. The access aims to help participating organizations strengthen digital infrastructure by identifying security weaknesses and accelerating patches. However, major Indian IT services companies remain excluded from the current list of beneficiaries.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.