La preview de Claude Mythos de Anthropic señala un deshielo en las relaciones con el gobierno de EE. UU.

Key Points

• Anthropic’s Claude Mythos Preview is a cybersecurity‑focused AI model designed to detect vulnerabilities in major software.
• The model is already being used by Apple, Nvidia and JPMorgan Chase to preempt attacks.
• After months of conflict with the Trump administration, Anthropic is in talks with senior U.S. officials about Mythos’s capabilities.
• CEO Dario Amodei reportedly scheduled a meeting with White House chief of staff Susie Wiles.
• Ballard Partners, a lobbying firm linked to Trump, has been hired to facilitate government outreach.
• CISA and parts of the intelligence community are testing Mythos Preview; other agencies have shown interest.
• Anthropic previously cleared its models for classified DoD networks, highlighting its existing government ties.