Project Glasswing: How India’s Digital Frontier Gains Access to Anthropic’s Mythos AI

In a major boost to national cyber-defenses, key Indian government agencies are securing access to Anthropic's powerful, unreleased Mythos model to fortify critical infrastructure.

India’s cybersecurity landscape is bracing for a significant technological shift as the government secures access to "Project Glasswing," a high-stakes initiative led by US-based firm Anthropic. Through this programme, several of the country’s most sensitive institutions are gaining access to the Mythos AI model—a closely guarded, unreleased technology designed specifically to detect software vulnerabilities at a pace and scale that could shift the advantage from attackers to defenders.

Government entities, including the Indian Computer Emergency Response Team (CERT-In), the Indian Cyber Crime Coordination Centre (I4C), and the National Critical Information Infrastructure Protection Centre (NCIIPC), are among the primary beneficiaries. Sources indicate that these agencies, which report to the Prime Minister’s Office and the Ministry of Electronics and Information Technology, specifically sought this access to stress-test the security protocols of India’s banking, power, and telecommunications grids.

Scaling the Global Defense Effort

The move marks a strategic expansion of Project Glasswing, which is now being rolled out across 15 nations. While the model was initially restricted to a narrow cohort in the US and UK due to concerns regarding the potential for misuse, Anthropic is now broadening the reach of the Mythos Preview. This international expansion includes the European Union’s cybersecurity agency, ENISA, signaling that the model is being positioned as a global standard for proactive threat hunting.

The decision to share such a potent tool comes amidst a backdrop of "cybersecurity hysteria" in global markets. While some industry analysts have raised alarms that the model’s capabilities could inadvertently expose weaknesses in nearly every major computer system on Earth, practitioners who have already gained access to the platform suggest the true value lies in the speed of response. For these teams, the model’s ability to streamline the discovery of vulnerabilities is far more critical than the theoretical risks associated with its misuse.

Balancing Innovation and Exclusion

Despite the excitement, the rollout has not been without friction. Reports from the global tech sector suggest that while government agencies are gaining access, some private-sector operational technology providers have expressed annoyance at being excluded from the early stages of the Mythos deployment. Furthermore, the landscape remains competitive; with UK banking institutions reportedly turning to OpenAI’s GPT-5.5, it is clear that nations and corporations are diversifying their reliance on various frontier models to ensure national digital security.

Within India, the government’s efforts extend beyond state agencies. Discussions are currently underway to provide select cybersecurity and research teams within the country’s leading IT services firms with access to the model. This public-private collaboration aims to create a more resilient digital ecosystem, ensuring that the software powering India’s most essential services remains one step ahead of increasingly sophisticated global cyber threats.