Anthropic leak rattles markets — “Claude Mythos” details exposed, security stocks and bitcoin slide
Anthropic, the AI firm behind Claude, has begun limited testing of a new model it calls “a step change” in performance — “the most capable we’ve built to date,” the company says — but an accidental data exposure overnight has amplified investor and cybersecurity concerns.
What leaked
According to Fortune, around 3,000 assets tied to Anthropic’s blog were left in a publicly accessible data store, exposing internal drafts and unreleased materials. Among them was a draft blog post that used the name “Claude Mythos” for the new model and warned that the system could pose “serious cybersecurity risks,” including the ability to identify and potentially exploit software vulnerabilities.
What Anthropic is building
Anthropic currently offers three model tiers — Haiku, Sonnet and Opus — and is testing the new, more capable model with a small group of early access customers while it evaluates behavior and risks. The leaked files also suggest the company is developing an even larger model tier, dubbed “Capybara,” positioned above Opus in size and capability.
Market impact
The leak hit cybersecurity-focused software stocks and tech-software ETFs. Palo Alto Networks (PANW), CrowdStrike (CRWD) and Fortinet (FTNT) each slid roughly 4%–6%, and the iShares Expanded Tech-Software Sector ETF (IGV) fell about 2.5%.
Crypto reaction
The overnight disclosure also coincided with a pullback in bitcoin: after flirting with $70,000 earlier in the day, BTC dropped back toward $66,000 — a decline of roughly 5–6% — a move traders say the leak likely helped trigger as risk sentiment shifted.
Why it matters for crypto and security
The episode underscores two immediate issues for crypto markets and digital infrastructure: rapid advances in AI can change the threat landscape for software vulnerabilities, and accidental disclosures of internal AI development materials can quickly ripple through both equity and crypto markets. Investors and security teams will be watching how Anthropic manages model rollout, containment, and mitigation of the risks flagged in the leaked documents.
Read more AI-generated news on: undefined/news
