# UK financial regulators are holding critical talks with the government’s cybersecurity agency and major banks to assess risks posed by the latest artificial intelligence model.

*genai · news · 2026-04-12 · Firstpost*

## Key points

- UK regulators and the National Cyber Security Centre are assessing risks from Anthropic's Claude Mythos Preview AI.
- Claude Mythos Preview has uncovered thousands of vulnerabilities, including critical flaws in the Linux kernel.
- Anthropic is restricting access to Claude Mythos Preview under Project Glasswing nL4b40Q0LK for defensive use only.
- Major banks, insurers, and exchanges will be briefed on the model's covert security risks within two weeks.
- Anthropic is offering the model to select companies like Apple, Microsoft, and Google before public release.

**Companies:** Apple, Microsoft, Google, Anthropic
**Countries:** UK, US

[Read the full story on Firstpost](https://www.firstpost.com/tech/uk-financial-authorities-assess-potential-risks-of-anthropics-new-ai-model-report-13999579.html)

---

Canonical: https://newsio.io/n/978fe62f-6236-47d1-a445-d725f73529b5/uk-financial-regulators-are-holding-critical-talks-with-the-governments-cybersec
Summarized by Newsio from Firstpost. https://newsio.io/how-it-works