Several Indian organisations from the cybersecurity, telecom, banking and finance sectors have received early access to Claude Mythos, the highly restricted artificial intelligence model developed by US-based AI company Anthropic. Senior government officials confirmed the development on Wednesday, a day after Anthropic announced a major expansion of its Project Glasswing cybersecurity initiative.

According to officials, the number of Indian organisations that have gained access to the model remains in single digits. However, the government is working to expand participation. Notably, major Indian information technology services companies are not part of the current list of organisations that have been granted access.

"The government is actively engaging with Anthropic to widen the access further," a senior official told ET.

The official added that the Indian government will receive a preview of the Mythos model through selected public sector entities. These organisations could also include the Indian Computer Emergency Response Team (CERT-In), the country's nodal cybersecurity agency. The government had specifically requested access for CERT-In during discussions with Anthropic.

Officials declined to reveal the names of the Indian organisations that have been included in the latest batch. One official said, "Anthropic has requested the 150 additional entities to not discuss the details for a couple of days. The company will officially announce the names in the coming days."

The development follows months of discussions between the Indian government and Anthropic. Earlier, ET had reported that the Ministry of Electronics and Information Technology (MeitY) had been in talks with the US administration for nearly two months. The objective was to ensure India gained access to Mythos and that the country's critical infrastructure remained protected against potential cyber threats.

Anthropic first rolled out access to Mythos in April to a select group of 50 organisations. The initial list included major cloud service providers such as Amazon Web Services, Google Cloud and Microsoft Azure. Technology companies, including Nvidia and Apple were also granted access. Among financial institutions, JPMorgan Chase was part of the first cohort.

According to Anthropic, organisations that received early access have already used the model to identify more than 10,000 previously unknown high-severity and critical security vulnerabilities. The company says these discoveries have helped improve cybersecurity preparedness across several sectors.

On Tuesday, Anthropic announced that it would expand the Glasswing partnership programme to include 150 additional organisations from more than 15 countries. The decision came after weeks of collaboration with existing partners, cybersecurity experts, open-source software maintainers and agencies of the US government.

The company stated that every organisation selected for the programme must meet strict security standards before gaining access to the Mythos model. Anthropic considers the model highly sensitive due to its advanced capabilities and potential applications in cybersecurity.

The newly expanded group includes sectors that were not adequately represented in the first round of participants. These sectors include power, water, healthcare, communications and hardware. Several of the new organisations are vendors and non-profit entities that maintain software code used by numerous companies and government agencies around the world. Their inclusion is expected to strengthen cybersecurity across a wider range of critical infrastructure systems.

The Indian government has been closely monitoring the possible impact of the Mythos model on domestic systems and databases. Over the past month, Finance Minister Nirmala Sitharaman, Electronics and IT Minister Ashwini Vaishnaw, senior officials from their ministries and regulators such as the Reserve Bank of India have held multiple meetings with industry stakeholders. These discussions focused on understanding the opportunities and risks associated with the advanced AI model.

Many Indian companies have shown interest in securing access to Mythos. However, government officials had earlier clarified that the Centre would not directly represent private companies seeking access. Instead, the government's engagement with Anthropic has been focused on national concerns and critical infrastructure protection.

MeitY Secretary S Krishnan had previously stated that the government was interacting directly with Anthropic and had communicated its concerns regarding cybersecurity and access to advanced AI technologies.

Indian IT services companies have also been eager to gain access to the Mythos model. Industry body Nasscom has held discussions with Anthropic regarding the participation of Indian technology firms in the programme. According to industry representatives, access to Mythos could significantly strengthen cyber resilience because enterprises worldwide depend on software and code developed by Indian IT companies. The model is widely regarded as one of the most powerful AI systems available today, making it a valuable tool for identifying security weaknesses before they can be exploited.

Queries sent by ET to Anthropic, Nasscom and leading Indian IT services companies including Tata Consultancy Services, Infosys, HCLTech, Wipro and Tech Mahindra did not receive responses until the time of publication.