Govt's cybersecurity efforts get Anthropic's Claude Mythos boost

Central agencies like I4C and Cert-In, as well as some IT firms, to get limited access for AI model testing and deployment

The Indian Cyber Crime Coordination Centre (I4C), the Indian Computer Emergency Response Team (Cert-In), the National Critical Information Infrastructure Protection Centre (NCIIPC), and the Digital Intelligence Platform (DIP) of the Department of Telecommunications are among a handful of government agencies that are likely to have been provided limited access to Claude Mythos, according to sources. The access to the AI model, initially for testing and later for full-scale deployment, will help identify vulnerabilities in the critical digital infrastructure of the central and state governments, the sources said. 

Apart from these government agencies, some Indian companies in the information technology (IT) and information technology-enabled services (ITes) sector may also have been provided limited access to Claude Mythos, “provided they form small teams with engineers and researchers who have been doing cutting-edge work on AI (artificial intelligence)”, a source said. An email sent to Anthropic did not elicit any response. 

Claude Mythos is Anthropic's latest generative AI model. While not designed for cybersecurity tasks, the general purpose large language model has demonstrated exceedingly strong capabilities in locating dormant bugs and identifying loopholes, as well as legacy exploits in old and very old software systems. I4C, which is housed under the Ministry of Home Affairs (MHA), functions as the government’s nodal agency to fight cybercrime, while Cert-In, housed under the Ministry of Electronics and Information Technology (Meity), is the nodal agency responsible for cyber and digital security of the Indian internet. 

The NCIIPC, which functions under the National Technical Research Organisation of the National Security Advisor’s office, primarily works on safeguarding power stations, including nuclear power plants, from digital and cyber threats. On the other hand, the DIP, which operates under the DoT, functions as the inter-agency coordination body for telecom-related fraud prevention. 

These companies and government agencies have been provided access to Claude Mythos under Anthropic’s Project Glasswing. Earlier this week, the Dario Amodei-led company announced that it was expanding its cyberdefence plan, Project Glasswing, to 150 new organisations across 15 countries. 

“The group covers several industries that weren’t well represented in our initial cohort, such as power, water, health care, communications, and hardware. And many of the new partners are vendors — companies or nonprofits that maintain codebases that are relied upon by lots of other organisations around the world, including governments,” Anthropic said. 

The move to expand Claude Mythos and Project Glasswing came just weeks after the domestic software industry body Nasscom, as well as Meity, held meetings with Anthropic representatives, seeking access for Indian government agencies and companies. 

Earlier this year, in April, Anthropic joined hands with leading global companies, such as Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks, to find solutions to deploy Claude Mythos’s prowess “for defensive purposes”. 

Project Glasswing was announced shortly after the release of Claude Mythos. While announcing Project Glasswing, Anthropic had said that Mythos Preview had already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser. 

In its press note announcing the expansion of Project Glasswing, Anthropic said that the capabilities of the Claude Mythos preview have sparked new discussions about the impact AI is likely to have on cybersecurity. 

“Cheap, fast AI models with powerful cyber capabilities are around the corner. We want Project Glasswing to spur institutions toward operating norms that reflect this reality,” the company said.