# Anthropic has released the first AI model capable of hacking IT networks.

*genai · news · 2026-04-13 · The Telegraph*

## Key points

- Anthropic's Claude Mythos is the first AI model capable of autonomously executing advanced cyberattacks.
- AISI testing showed Mythos can perform multi-step hacks in hours that would take humans days.
- Anthropic has withheld public release of Mythos due to its unprecedented hacking capabilities.
- Mythos has already identified thousands of vulnerabilities in widely used browsers and operating systems.
- Major financial institutions and regulators are holding emergency meetings to address Mythos-related cybersecurity threats.

**Companies:** Google, Nvidia, JP Morgan, Anthropic
**Countries:** United Kingdom, United States

[Read the full story on The Telegraph](https://www.telegraph.co.uk/business/2026/04/13/anthropic-model-is-first-ai-to-hack-networks/)

---

Canonical: https://newsio.io/n/a59b89f0-02c4-4f55-b08d-f15b0147d58e/anthropic-has-released-the-first-ai-model-capable-of-hacking-it-networks-uk-safe
Summarized by Newsio from The Telegraph. https://newsio.io/how-it-works