Anthropic released Claude Fable 5 publicly on Tuesday, rattling crypto traders who fear its capabilities could accelerate the exploits behind more than $600 million in recent thefts.
Key Points:
- Anthropic launched Claude Fable 5, its most capable public model, with guardrails that route risky cybersecurity prompts to its older Opus 4.8.
- Crypto founders warned the tool could speed up attacks during a wave of recent thefts across the industry.
- A less restricted version, Mythos 5, went to Project Glasswing partners instead of the general public.
Claude Fable 5 Launches With Built-In Guardrails
The company pushed the model out Tuesday afternoon, handing the public its first access to Anthropic's most powerful tier. Claude Fable 5 runs on the same technology as the firm's restricted Mythos system, which alarmed the cybersecurity world earlier this year. Anthropic said the model outperforms anything it has ever offered for general use, and priced it at roughly twice the cost of Claude Opus, its prior flagship.
To blunt misuse, the firm diverts sensitive prompts on cybersecurity, biology and chemistry to its older Opus 4.8 model. It opened Fable 5 free to Pro, Max, Team and Enterprise subscribers until Jun. 22, then moves them onto paid compute credits.
Also Read: OpenAI Joins $852B IPO Race As Anthropic And SpaceX Charge Ahead
Crypto Founders Warn Of Online Danger
The launch still unnerved parts of the crypto market, where several founders braced for turmoil ahead of a rollout prediction markets saw as near certain. Floors Finance CEO Omer Demirel said the release risked a "doomsday for the internet." Intuition CEO Billy Luedtke expected interacting with almost anything online to feel "dangerous and scary for a bit."
Luedtke still predicted a stronger security era for crypto once the early panic and market chop faded. Analyst The DeFi Investor urged users to revoke token approvals, lean on heavily audited apps and split holdings across several wallets to cut single points of failure.
Curve Finance founder Michael Egorov played down the threat to decentralized finance itself, calling much of the alarm overblown. He argued the model's flair for spotting browser bugs does not translate cleanly to deployed smart contract code.
Egorov expects attackers to chase operational weak spots instead, such as compromised multisig keys and tampered frontend dependencies.
Mythos Fears Trace Back To April
Market commentator The Milk Road warned of short-lived FUD while rating the chance of a serious hack as low. It said front-ends, bridges and the servers guarding private keys stay softer targets than contracts that big teams audit relentlessly.
Anthropic first revealed the Mythos technology in April through Project Glasswing, sharing it with 50 to 60 vetted firms and later more than 150 organizations. Its researchers concluded that one operator using such a model could turn a month of patches into working exploits in a single afternoon for just a few thousand dollars. The release lands during a punishing stretch for crypto security, with attackers draining more than $600 million since early April.
Read Next: Bitcoin Drawdown Reveals Why Big Buyers Still Want More BTC