AI can now build an OS from scratch, claims Anthropic
In just eight days, Anthropic's Claude Fable 5.2 created a functional prototype operating system. This involved more than 1,000 dedicated AI agents who tirelessly collaborated to craft and evaluate the code.
By ET AI |
ET AI
Anthropic's AI builds an operating system using 1,000 AI agents (AI-generated image)
Artificial intelligence (AI) can now write code to build software systems, with Anthropic claiming that its latest model, Claude Fable 5.2, has built a working operating system (OS) from scratch in eight days using over 1,000 AI agents.
“As an experiment, one Anthropic engineer gave Fable 5.2 the task of designing a prototype operating system. Starting from an empty folder, the model coordinated over 1,000 subagents, working around the clock for eight days writing, testing, and reviewing code,” the company said in a blog post on Tuesday.
The post added: “It produced a working operating system that boots on a laptop, with its own core that runs the machine, drivers for the disk, keyboard, and network, permanent file storage, windows, a login screen, an installer, and built-in Claude functionality,”
Anthropic claimed Fable 5.2 showed strong engineering judgement and could produce high-quality results even from ambiguous requests. “Over 90% of the changes merged into Anthropic’s core codebase are now written by Claude (with human guidance),” it said.
The model was used to fix stubborn bugs, optimise performance across Anthropic’s services and complete projects that would otherwise have been too slow or expensive to attempt.
The OS experiment is part of a broader set of demonstrations highlighting the growing capabilities of multi-agent AI systems. In another example, a team of 1,000 Fable 5.2 agents formalised the proof of Fermat's Last Theorem in 30 hours, with minimal human intervention.
Mathematical formalisation involves translating an existing proof into a language that a computer can verify step by step. Anthropic said an earlier, unreleased research version of Claude had taken 11 days to complete the same exercise, relying on a purpose-built system and guidance from a human researcher.
By comparison, Fable 5.2 started with minimal prompts, developed its own coordination methods and used no mathematical references; instead, it relied on intermediate theorem statements to track progress. The company said the model used a third of the action tokens required in the earlier attempt and produced a proof seven times shorter.
“As an experiment, one Anthropic engineer gave Fable 5.2 the task of designing a prototype operating system. Starting from an empty folder, the model coordinated over 1,000 subagents, working around the clock for eight days writing, testing, and reviewing code,” the company said in a blog post on Tuesday.
The post added: “It produced a working operating system that boots on a laptop, with its own core that runs the machine, drivers for the disk, keyboard, and network, permanent file storage, windows, a login screen, an installer, and built-in Claude functionality,”
Anthropic claimed Fable 5.2 showed strong engineering judgement and could produce high-quality results even from ambiguous requests. “Over 90% of the changes merged into Anthropic’s core codebase are now written by Claude (with human guidance),” it said.
The model was used to fix stubborn bugs, optimise performance across Anthropic’s services and complete projects that would otherwise have been too slow or expensive to attempt.
The OS experiment is part of a broader set of demonstrations highlighting the growing capabilities of multi-agent AI systems. In another example, a team of 1,000 Fable 5.2 agents formalised the proof of Fermat's Last Theorem in 30 hours, with minimal human intervention.
ADVERTISEMENT
Mathematical formalisation involves translating an existing proof into a language that a computer can verify step by step. Anthropic said an earlier, unreleased research version of Claude had taken 11 days to complete the same exercise, relying on a purpose-built system and guidance from a human researcher.
By comparison, Fable 5.2 started with minimal prompts, developed its own coordination methods and used no mathematical references; instead, it relied on intermediate theorem statements to track progress. The company said the model used a third of the action tokens required in the earlier attempt and produced a proof seven times shorter.
Download
The Economic Times Business News App for the Latest News in Business, Sensex, Stock Market Updates & More.
The Economic Times Business News App for the Latest News in Business, Sensex, Stock Market Updates & More.
Related Articles
Anthropic unveils Fable 5.2 and Mythos 5.2, its most powerful AI models yet: All you need to knowADVERTISEMENT
