Anthropic Restricts Mythos to Dozens of Testers as CEO Cuts Europe Trip Short
Updated
Updated · Bloomberg · May 28
Anthropic Restricts Mythos to Dozens of Testers as CEO Cuts Europe Trip Short
4 articles · Updated · Bloomberg · May 28
Anthropic has limited its new Mythos AI model to a few dozen organizations for testing and preview, reflecting internal concern that the system is too dangerous for a full public launch.
Dario Amodei cut a planned European trip to one or two days in the UK to manage the rollout, after previously signaling last month that Mythos was too risky to release.
Staff with access say Mythos feels more human than earlier models, but Anthropic worries the same capabilities could be weaponized for cyberattacks.
The launch matters beyond product timing: Mythos is central to Amodei’s effort to improve Anthropic’s standing with the Trump administration while preparing for a broader public debut.
A private firm built an AI 'too dangerous to launch.' Who should decide who can wield this unprecedented power?
As AI learns to hack better than humans, is the internet's fundamental security already broken?
Anthropic’s Claude Mythos: How Restricted AI Access Is Reshaping Global Cybersecurity and Geopolitics
Overview
Anthropic’s decision to restrict access to its advanced AI model, Claude Mythos, has sparked major controversy, especially in Europe. While top U.S. banks like JPMorganChase and Citigroup are testing Mythos, European banks—who rely on aging software—are left out and increasingly worried about the model’s ability to rapidly find software vulnerabilities. This exclusion highlights a growing geopolitical divide, as access to powerful AI models becomes a strategic issue. The situation raises fears of cascading failures in Europe’s finance sector and underscores the urgent need for international cooperation and stronger cybersecurity measures.