Anthropic Confirms Major Source Code Leak of Claude AI
Anthropic, the AI unicorn, recently found itself at the center of an unprecedented public relations crisis. Developer Chaofan Shou revealed on a social platform that a critical oversight during an npm package publication left crucial .map files intact. This error resulted in the complete source code exposure of its latest command-line tool, Claude Code .

The scale of the leak is staggering, encompassing nearly two thousand files and over 500,000 lines of TypeScript code. Despite the official team's emergency response, the source code was swiftly archived on GitHub and has been permanently circulated within the developer community. This fundamental mistake laid bare Anthropic's technical secrets to competitors.
Romantic and Hardcore: Code Reveals "Cyber Pet" and "Nocturnal Dreaming" Features
While the leak was deeply embarrassing, the unreleased features hidden within the code have captivated the industry. A project codenamed BUDDY has drawn particular interest. It can generate unique, pixel-style cyber pets based on user IDs to accompany developers while coding, complete with personalized traits like a configurable "sarcasm level."
Even more ambitious is a function codenamed KAIROS, described as an "always-on Claude." The source code indicates it incorporates a bionic "nocturnal dreaming" mechanism. This system works in the background to process the day's interaction fragments, removing redundancies and consolidating core content to train the AI into a more understanding, long-term companion for the user.
This disastrous "wild launch" has ignited profound discussions about AI safety and governance. For a company that prides itself on being "safe and cautious," Anthropic's "intern-level" mistake exposes potential fragility in the foundational management of leading tech firms as they race for R&D velocity.
The exposure of these 500,000 lines of code represents more than a loss of intellectual property; it serves as a collective warning to the entire AI industry. As AI agents are granted greater autonomy to manage code and systems, even a minor engineering oversight can escalate into a globally disruptive event.
Related article
DeepL, renowned for text translation, now targets voice translation
DeepL, a translation company best known for its text-based tools, has launched a voice-to-voice translation suite today that addresses scenarios such as meetings, mobile and web conversations, and group discussions for frontline workers through custo
Talat’s AI meeting notes live on your device, not the cloud
Granola, the AI-powered notetaking app valued at $250 million, has gained traction among tech founders and venture capitalists. But one developer sees demand for a more private, fully local alternative available for a one-time fee with no subscriptio
New Roewe i6 Hits Market at 659,000 Yuan, Powered by Snapdragon 8155 and Doubao Large Model
SAIC Roewe today launched the new Roewe i6, a compact sedan that fully adopts the visual language of the Roewe D7. Its distinctive large upright grille and horizontal halo light bar stretch across the front, creating a strong sense of technology and
Related Special Topic Recommendations
Comments (0)
0/500
Anthropic, the AI unicorn, recently found itself at the center of an unprecedented public relations crisis. Developer Chaofan Shou revealed on a social platform that a critical oversight during an npm package publication left crucial .map files intact. This error resulted in the complete source code exposure of its latest command-line tool,

The scale of the leak is staggering, encompassing nearly two thousand files and over 500,000 lines of TypeScript code. Despite the official team's emergency response, the source code was swiftly archived on GitHub and has been permanently circulated within the developer community. This fundamental mistake laid bare Anthropic's technical secrets to competitors.
Romantic and Hardcore: Code Reveals "Cyber Pet" and "Nocturnal Dreaming" Features
While the leak was deeply embarrassing, the unreleased features hidden within the code have captivated the industry. A project codenamed BUDDY has drawn particular interest. It can generate unique, pixel-style cyber pets based on user IDs to accompany developers while coding, complete with personalized traits like a configurable "sarcasm level."
Even more ambitious is a function codenamed KAIROS, described as an "always-on Claude." The source code indicates it incorporates a bionic "nocturnal dreaming" mechanism. This system works in the background to process the day's interaction fragments, removing redundancies and consolidating core content to train the AI into a more understanding, long-term companion for the user.
This disastrous "wild launch" has ignited profound discussions about AI safety and governance. For a company that prides itself on being "safe and cautious," Anthropic's "intern-level" mistake exposes potential fragility in the foundational management of leading tech firms as they race for R&D velocity.
The exposure of these 500,000 lines of code represents more than a loss of intellectual property; it serves as a collective warning to the entire AI industry. As AI agents are granted greater autonomy to manage code and systems, even a minor engineering oversight can escalate into a globally disruptive event.
DeepL, renowned for text translation, now targets voice translation
DeepL, a translation company best known for its text-based tools, has launched a voice-to-voice translation suite today that addresses scenarios such as meetings, mobile and web conversations, and group discussions for frontline workers through custo
Talat’s AI meeting notes live on your device, not the cloud
Granola, the AI-powered notetaking app valued at $250 million, has gained traction among tech founders and venture capitalists. But one developer sees demand for a more private, fully local alternative available for a one-time fee with no subscriptio
New Roewe i6 Hits Market at 659,000 Yuan, Powered by Snapdragon 8155 and Doubao Large Model
SAIC Roewe today launched the new Roewe i6, a compact sedan that fully adopts the visual language of the Roewe D7. Its distinctive large upright grille and horizontal halo light bar stretch across the front, creating a strong sense of technology and





Home






