RBI Said to Evaluate Cybersecurity Risks Linked to Anthropic’s Mythos Amid Concerns of Unauthorised Access

Anthropic’s new artificial intelligence (AI) model, Mythos is said to be a “regular” bank of India (RBI), which talks with international regulators, domestic lenders and government officials about potential risks associated with Antroppic. RBI’s preliminary assessment points towards Mythos may raise cybersecurity concerns by speeding up the discovery and use of software vulnerabilities, a report said. The move follows the revelations of unauthorised personnel gaining access to Anthropic’s Mythos, which is described as ‘so powerful that it could lead to dangerous cyberattacks’.

RBI Evaluating Mythos-Linked Cybersecurity Risks

The RBI, in the last two weeks, has consulted with counterparts around the world, according to sources familiar with the matter. In addition, this reportedly includes the Federal Reserve and also the Bank of England, which is hoping to understand what are new threats or protections will be.

Claude Mythos Discussion

• Claude Mythos Preview has scared me. Can AI really hack websites and social media accounts?
• Claude “Mythos leak on Discord! How did they get in?
• Anthropic’s Claude Mythos Preview is scary!

Explore More…

It quoted one source as saying, “Globally we are talking to other countries and other regulators about what is the developments and what safeguards should be taken.”

In the report, it says that RBI may also pursue direct relationship with Anthropic. Moreover, regulators in all parts of Asia, Europe and the US are said to have advised banks to review their cybersecurity preparedness. National Payments Corporation of India (NPCI), which provides payment services such as UPI in the country, is said to be exploring early access to Mythos with a small group of banks too.

Reuters said this is to identify possible “day-zero” vulnerabilities before any wider rollout, although such access could be restricted by restrictions on the use of a source. In the US, it is said that Mythos is hosted on tightly controlled servers under Mytos in a host country called tight control. Consequently, any tests on local datasets in foreign jurisdictions could be regulatory and technical challenges.

The RBI also is said to be working on more detailed guidelines for banks entering enterprise partnerships with advanced AI models such as Anthropic’s Mythos and Claude family. But if there was any deployment of Indian user data that would have to comply with RBI’s data localisation requirements, the paper said ‘A source is saying this.

Concerns Over Unauthorised Access to Mythos

Regulatory debates follow a short time after reports that ‘a small group of unauthorised users had early access to Mythos’. But the model, which Anthropic itself has described as “very powerful,” was also available through a private discord organization that came to be reported on the same day it was announced for limited testing, Bloomberg said.

Although it is said that the group did not use the model for malicious purposes, the incident has raised concerns about possible misuse of the. The group shared screenshots that showed a Mythos dashboard at the time. This includes the panel user management panels, AI experiment interfaces and detailed analytics for model performance and cost.

A thropic is now investigating the matter. The company has said in a statement that we’re investigating an investigation of claiming unauthorised access to Claude Mythos Preview through one of our third-party vendor environments.