Detailed Analysis
India's central government is reportedly in preliminary discussions with both the United States government and Anthropic to secure access to Claude Mythos, Anthropic's frontier AI model distinguished by its advanced autonomous cybersecurity capabilities, according to Inc42. The report surfaces at a particularly sensitive moment in the model's short public history: Mythos has already demonstrated the ability to independently identify and exploit thousands of software vulnerabilities, prompting Anthropic CEO Dario Amodei to hold a meeting at the White House on April 17, 2026, with senior officials including Chief of Staff Susie Wiles. That meeting resulted in Anthropic dialing back the cybersecurity features of its publicly available model, Claude Opus 4.7, reflecting the depth of national security concern the model's capabilities have generated within the US government.
Rather than pursue a broad commercial release, Anthropic responded to those concerns by launching Project Glasswing, a tightly controlled preview program granting access to approximately 40 vetted organizations. Current participants include major technology and financial firms such as Amazon, Apple, Microsoft, Google, Cisco, CrowdStrike, Nvidia, JPMorgan Chase, and Broadcom, with use cases restricted to defensive cybersecurity applications — specifically the protection of critical infrastructure and the hardening of open-source code. Anthropic has committed $100 million in credits to the initiative, and post-preview pricing is set at $25 per million input tokens and $125 per million output tokens, available through the Claude API, Amazon Bedrock, Vertex AI, and Microsoft Foundry. The structured rollout signals a deliberate strategy to extract defensive value from Mythos while preventing offensive misuse.
If the Inc42 report is accurate, India's interest in Mythos access would represent a significant geopolitical dimension to Project Glasswing's expansion. As one of the world's largest digital economies with extensive critical infrastructure increasingly exposed to sophisticated cyber threats, India has clear strategic motivation to seek access to state-of-the-art defensive AI tooling. The bilateral framing of the reported talks — involving both the US government and Anthropic directly — suggests that access to Mythos is being treated not merely as a commercial transaction but as a matter of allied technology diplomacy, consistent with the broader US posture of managing Mythos as a dual-use national security asset rather than a standard software product.
The episode underscores a broader and accelerating tension in frontier AI development between commercial deployment imperatives and national security constraints. Anthropic's handling of Mythos — restricting public access, convening White House-level discussions, and constructing a vetted consortium model — sets a precedent for how advanced AI capabilities that approach or cross into weapons-relevant territory may be governed going forward. The reported Indian government interest further illustrates how such decisions are rapidly acquiring a foreign policy dimension, with allied governments positioning to ensure they are not left outside the access perimeter as AI-enabled cyber capabilities become a defining element of national security infrastructure. Whether India's reported talks translate into formal Glasswing participation or a separate bilateral arrangement will be closely watched as a signal of how broadly the US and Anthropic intend to share access to Mythos-class capabilities beyond domestic partners.
Read original article →