Anthropic Claims Its New A.I. Model, Mythos, Is a Cybersecurity ‘Reckoning’
The New York Times • • business
Key Points:
- Anthropic has developed a new AI model called Claude Mythos Preview, which it considers too powerful for public release due to potential risks.
- Instead of a public launch, Anthropic will provide access to a consortium of over 40 tech companies, including Apple, Amazon, and Microsoft, to help identify and fix security vulnerabilities in critical software.
- The coalition, named Project Glasswing, also includes competitors like Google, hardware providers such as Cisco and Broadcom, and organizations like the Linux Foundation, with Anthropic committing up to $100 million in usage credits.
- Anthropic aims to raise awareness about emerging AI threats and accelerate efforts by "good actors" to secure both open-source and private infrastructure against these risks.
- Company leaders describe this initiative as a pivotal moment for the AI industry to address and manage the dangerous capabilities of advanced models.
 { path { fill: %23EEEFE9 } })
