5 Key Facts About Anthropic's Project Glasswing To Secure 'World's Most Important Software'

Published 7 June 2026 · tech

5 Key Facts About Anthropic's Project Glasswing To Secure 'World's Most Important Software' Published By, Powered by Last Updated: June 07, 2026, 11:40 IST Anthropic

5 Key Facts About Anthropic's Project Glasswing To Secure 'World's Most Important Software' Published By, Powered by Last Updated: June 07, 2026, 11:40 IST Anthropic has extended access to its closely guarded cybersecurity initiative called Project Glasswing to more than 15 countries, including India. Rapid Read Anthropic has expanded access to its restricted Project Glasswing initiative. (File) US artificial intelligence company Anthropic has extended access to its closely guarded cybersecurity initiative called Project Glasswing to more than 15 countries, including key government entities in India. Anthropic has described Project Glasswing as a “collaborative effort to secure the world’s most important software". The primary function of the project is to identify critical software vulnerabilities at superhuman speeds, marking an unprecedented push in global cybersecurity efforts. A handful of Indian entities from both India’s government and private sectors have been granted access to Claude Mythos Preview, an advanced AI model built to identify weaknesses in software systems before they can be exploited.

5 Facts About Project Glasswing Anthropic launched a highly restricted cybersecurity initiative called ‘Project Glasswing’ designed to protect critical software infrastructure using an advanced AI model. The project brings together major technology firms, security researchers, governments, and open-source maintainers to identify and fix vulnerabilities before malicious actors can exploit them. At the centre of the project lies Claude Mythos Preview, the latest AI model by Anthropic that is believed to have exceptional coding and cybersecurity capabilities. Anthropic and early partners said Mythos can scan vast amounts of code and “chain together" security flaws at a speed unmatched by any human hacker, raising fears that it can be used to bring down critical national infrastructure in a matter of hours. The organisation involves some of the world’s biggest tech players and verified partners such as Amazon, Apple, Microsoft, Google, Cisco, CrowdStrike, and JPMorgan Chase.

Anthropic says these organisations are using the model to strengthen the security of critical infrastructure and software systems. Anthropic says Claude Mythos Preview has identified more than 10,000 high and critical-severity security vulnerabilities, including flaws in major operating systems. However, some industry experts have said fears of unfettered hacking have been overstated. Earlier this month, Anthropic announced that it would expand Project Glasswing to at least 150 organisations in more than 15 countries, including Australia, Canada, France, Germany, Italy and India, provided they meet certain security requirements. Around 50 organisations were initially given access in April. News18 Newsletter Handpicked stories, in your inbox A newsletter with the best of our journalism submit About the Author Aveek Banerjee Aveek Banerjee is a Senior Sub Editor at News18. Based in Noida with a Master's in Global Studies, Aveek has more than three years of experience in digital media and news curation, specialising in int...