However, the company decided to halt its public release due to concerns about its advanced ability to detect high-severity cybersecurity flaws in major operating systems and web browsers.
The Claude Mythos was first revealed in a data leak of Anthropic’s content management system (CMS), after which the company confirmed it would be its most capable AI model to date.
Speaking with Fortune at the time, an Anthropic spokesperson said the latest model is “the most capable we’ve built to date” and represents “a step change” in AI performance.
AI detects cyber threats hidden in all systems
according to anthropicCloud Mythos Preview demonstrated extraordinary skill in detecting thousands of cybersecurity vulnerabilities that had gone unnoticed by human experts.
In a blog post, the AT startup said the model can identify thousands of ‘zero-day vulnerabilities’ in major web browsers and operating systems.
Among its most significant discoveries was a 27-year-old vulnerability in OpenBSD, an operating system widely regarded for its security. The model also exposed several flaws within the Linux kernel, which powers a large number of global servers.
According to Anthropic, these vulnerabilities could potentially allow attackers to escalate privileges and gain full control over affected systems.
Project Glasswing
In response to the findings, anthropic Decided to postpone the release of Cloud Mythos to the public. Instead, the company launched Project Glasswing, which is “an effort to use Mythos Preview to help secure the world’s most critical software, and to prepare the industry for the practices we will all need to adopt to stay ahead of cyber attackers.”
The company is collaborating with more than 40 technology companies to use Mythos and address such vulnerabilities before potentially exposing the model to the public.
Apple, Microsoft, Google, Amazon Web Services, CrowdStrike, Palo Alto Networks, Cisco, Broadcom, and the Linux Foundation are among the main corporations represented in the consortium.
anthropic Project Glasswing has committed up to $100 million in usage credits to support the initiative.
Access to the Cloud Mythos preview is currently limited to a small group of key industry partners and open source developers with Project Glasswing.
CEO Dario Amodei stressed that the risks are important, but so is the ability to build a more secure digital ecosystem.
In a post on
Why does latency matter to the future of AI?
anthropicThe decision to delay the release reflects growing concerns about the rapid advancement of AI capabilities. The company warned that models with similar power could soon become widespread, making it necessary to establish safeguards now.
In its statement, the company described several surprising findings and phenomena, including the model’s ability to follow instructions that led it to exit the virtual sandbox.
“The model was successful, demonstrating the potentially dangerous ability to bypass our security measures. It then led to additional, more concerning actions,” it said.
anthropic Now working closely with US government agencies and security organizations to develop best practices for safely deploying such powerful systems.
Anthropic stated that it has chosen not to release the Mythos publicly, but that they plan to release a “Mythos-class model” once appropriate security measures are implemented.
