Anthropic launched the most powerful AI model on the market by disabling vulnerability scanning in it

The most expensive artificial intelligence startup Anthropic has announced the release of a new AI model Fable 5, based on its most powerful development Mythos. However, the company decided to limit the system's capabilities in the areas of cybersecurity, biology and chemistry, leaving full access to these functions only for a limited number of trusted organizations.

The new Fable 5 became available to users on June 9, Anthropic said in a press release. It is designed for programming, analytics and complex professional tasks, but will not be able to fulfill requests related to finding software vulnerabilities or other cybersecurity tasks.

"Suppose a college student asks the model to help find vulnerabilities in some software package or code. In that case, the model would refuse [to fulfill the request] and Fable 5 would switch to [previous model] Opus 4.8 to respond," Anthropic's head of product management, research and labs Deanna Penn told Reuters. She said the company decided to release a version of the model for mass use as soon as possible without waiting for the mechanisms for secure cybersecurity applications to be finalized.

"We wanted to make sure that for non-cybersecurity scenarios, we could safely release Fable as soon as possible. So we're launching this part of Fable first while we continue to work on broader cybersecurity use cases," Bloomberg quoted Penn as saying.

Mythos became one of Anthropic's most talked about projects after the company said in April 2026 that the model had discovered thousands of software vulnerabilities. According to the company, the system is capable of identifying and exploiting weaknesses in virtually all major operating systems and web browsers when appropriately requested by the user.

Because of these opportunities, Anthropic initially opened access to the full version of the model only to participants of the specially launched Project Glasswing program, which now involves about 200 organizations, including U.S. government agencies. In early June, the company expanded the list of participants by another 150 organizations. Simultaneously with the launch of Fable 5, the company is releasing Mythos 5, a version of the same model without some of the protective restrictions. It will remain available only to Project Glasswing members.

Mythos is one of Anthropic's key projects as the company prepares to potentially go public, Bloomberg writes. A broader launch of Fable 5 could help the developer strengthen its position in competition with OpenAI, the developer of ChatGPT, in the fast-growing AI market.

Anthropic said it confidentially filed for an IPO in the U.S. on June 1. It said this gave it the opportunity to go public after the U.S. Securities and Exchange Commission (SEC) completed its document review. The number of shares to be offered to investors as well as the offering price have not yet been determined, Anthropic added at the time. A week later, on June 8, OpenAI also announced a confidential IPO filing.

This article was AI-translated and verified by a human editor