- Anthropic is granting early access to its powerful “Mythos” AI model to major tech firms like JPMorgan Chase and Microsoft.
- The crypto industry has been excluded from the “Project Glasswing” initiative, leaving it potentially vulnerable.
- Anthropic claims the model can find software vulnerabilities better than most human experts.
- Past tests showed even older AI agents could predict and find exploits in smart contracts.
On April 7, 2026, the $380 billion AI giant Anthropic announced a new frontier AI model called Mythos and a Cybersecurity initiative called Project Glasswing. However, it excluded the crypto industry from early access, potentially leaving billions in digital assets exposed.
Anthropic granted this head start to companies like JPMorgan Chase, Apple, and Microsoft. Consequently, crypto projects were not on the guest list for this paternalistic program, which aims to patch bugs before Mythos’s public release. One developer publicly asked Anthropic why it did not cooperate with crypto projects, but received no reply.
The company claims its unpublished model is “very autonomous” and has already found thousands of high-severity bugs. According to Anthropic, Mythos “can find software vulnerabilities better than all but the most skilled humans.”
Meanwhile, crypto’s software is seen as particularly vulnerable. Many smart contracts are immutable, hold immense value, and have slow upgrade cycles. In fact, previous tests in 2025 showed Anthropic’s AI agents could predict and find novel exploits in hundreds of contracts. Stifel analyst Adam Borg wrote the tool could elevate any Hacker to nation-state level.
✅ Follow BITNEWSBOT on Telegram, Facebook, LinkedIn, X.com, and Google News for instant updates.
