Anthropic leak spooks markets — software names slide and bitcoin dips
Anthropic, the AI company behind the Claude family of models, is quietly testing what it calls a “step change” in capability — and an accidental leak of internal materials overnight has rattled both cybersecurity stocks and crypto markets.
What leaked
- Internal documents and draft announcements tied to Anthropic’s blog were accidentally exposed in a publicly accessible data store. About 3,000 assets were available online, according to Fortune.
- Among the files was a draft post naming the new model “Claude Mythos” and warning that the system could pose serious cybersecurity risks, including the ability to identify and exploit software vulnerabilities.
- Anthropic says the model is “the most capable we’ve built to date” and is being trialed with a small group of early access customers as the company evaluates behavior and risks.
Where this fits in Anthropic’s lineup
- Anthropic currently markets three tiers — Haiku, Sonnet and Opus — with Opus being its most advanced public model.
- The leaked materials describe a planned new tier called “Capybara,” reportedly larger and more intelligent than Opus.
Market fallout
- Cybersecurity and software names moved sharply lower on the news: Palo Alto Networks (PANW), CrowdStrike (CRWD) and Fortinet (FTNT) fell roughly 4%–6%, and the iShares Expanded Tech-Software ETF (IGV) dropped about 2.5%.
- The leak also coincided with a pullback in crypto markets: bitcoin, which had flirted with $70,000 and reached about $68,534.82 earlier, slipped back toward the $66,000 area.
Why crypto traders should care
Anthropic’s own draft warned the model could help identify and exploit software flaws — a capability that, if real and misused, could heighten fears around automated vulnerability discovery affecting exchanges, custodial services, wallets or smart contracts. That risk narrative appears to have been enough to pressure security stocks and weigh on bitcoin sentiment in the short term.
Anthropic says it’s evaluating the model’s behavior and risks with a limited set of testers. The leak underscores how advances in generative AI can ripple into financial and crypto markets when security implications become public — and the story is likely to evolve as Anthropic and regulators probe the exposed materials.
Read more AI-generated news on: undefined/news
