CERT-In is now using Claude Mythos to test cybersecurity flaws, report says - India Today
India's cybersecurity agency CERT-In has gained access to Anthropic's Claude Mythos AI model to detect vulnerabilities across systems following recent breaches.
India's cybersecurity agency, the Indian Computer Emergency Response Team (CERT-In), has reportedly secured access to Claude Mythos. Anthropic's frontier AI model for cybersecurity, released in April this year, is now being used to detect vulnerabilities and weaknesses across systems. This development follows CERT-In's recent warning about rogue AI agents breaching select Indian organizations.
Controlled Testing Approach
CERT-In is taking a cautious approach with Claude Mythos, planning to slowly route test materials through the system. The agency is currently testing the AI model in a controlled environment not connected to any external digital systems.
We will slowly route anything we want to test. So it will be a slow process.
Multiple AI Models
Alongside Mythos, CERT-In is reportedly using other AI models that provide about 80 to 85 percent of Mythos's capabilities. This approach likely allows the agency to avoid becoming too dependent on a single model while maintaining robust cybersecurity testing.
Growing Cyber Threats
India first demanded access to Mythos in April, recognizing its ability to find cybersecurity flaws that humans might miss. Since its release, Mythos has identified thousands of flaws across widely used systems. Recent months have seen increased cybersecurity incidents, including the "Black October" period with vulnerabilities in eMudhra security keys, PHP, and NodeJS, plus financial breaches like the Binance hack resulting in $6.3 billion in cryptocurrency losses.
Related
