# Anthropic made a remarkable announcement about its new model, Claude Mythos Preview.

*genai · news · 2026-05-08 · The Guardian*

## Key points

- Anthropic restricted Claude Mythos Preview due to high operational costs, not just security concerns.
- OpenAI's GPT-5.5 and smaller models match Mythos in finding software vulnerabilities.
- Modern generative AI can autonomously discover and exploit vulnerabilities in both software and regulatory systems.
- Investment banks are already using AI to uncover previously unknown, complex tax loopholes at scale.
- Regulatory systems are much slower to patch vulnerabilities than software, making them uniquely vulnerable to AI-powered exploitation.

**Companies:** OpenAI, Anthropic, Aisle
**Countries:** UK, US

[Read the full story on The Guardian](https://www.theguardian.com/commentisfree/2026/may/08/how-dangerous-is-anthropics-mythos-ai)

---

Canonical: https://newsio.io/n/0d8bf3aa-fab5-4571-990e-ed1f585d36b9/anthropic-made-a-remarkable-announcement-about-its-new-model-claude-mythos-previ
Summarized by Newsio from The Guardian. https://newsio.io/how-it-works