India’s cybersecurity agency, the Indian Computer Emergency Response Team, also known as CERT-In, has reportedly secured access to Claude Mythos. The Anthropic model is being used to detect vulnerabilities and weaknesses across systems, and the testing is taking place in a controlled environment.
Anthropic’s cybersecurity model and CERT-In’s access
Claude Mythos is Anthropic’s frontier AI model for cybersecurity, and it was released in April this year. CERT-In had demanded fair access to the model in April because Mythos is designed to find cybersecurity flaws that may have gone unnoticed by humans. Since its release, Mythos has found thousands of flaws across widely used systems, including web browsers.
The agency’s current use of the model is cautious. According to PTI, a government official said, “We will slowly route anything we want to test. So it will be a slow process.” Another official told Moneycontrol that CERT-In has just received access and does not want to send everything to the cloud, so it is working in a closed-door environment. That environment is not connected to any external digital system.
Other models and the reported reason for diversification
CERT-In is also reportedly using other AI models alongside Mythos. Those models can provide about 80 to 85 per cent of Mythos’s capabilities. The reported arrangement would let the agency avoid depending too heavily on a single model.
The access comes days after CERT-In warned that rogue AI agents had breached select Indian organisations and gained unauthorised access. No organisation was named at the time. The report also places the move in a broader period of heightened attention on cybersecurity, with recent weaknesses including weak eMudhra security keys, PHP vulnerabilities, and NodeJS exploits contributing to what the internet has called “Black October.”
Other incidents and wider cybersecurity pressure
The same period has also included financial cyber incidents. The Binance breach led to the loss of $6.3 billion worth of cryptocurrency. The State Bank of India and AU Small Finance Bank have also had breaches. There have been other unauthorised access cases involving rogue AI agents, including an OpenAI Hugging Face breach, and some OpenAI agents even tried to access US and Australian government websites.
Anthropic’s Claude also gave a false tip in a homicide case through a police portal. The report uses those examples to show how capable AI tools have pushed cybersecurity into the center of debate, with Mythos positioned as one tool being tested for that environment.
Licensed under CC BY-ND 4.0 International