Anthropic News 2026: Leaks, Claude Mythos, and Explosive Growth (April 1–8, 2026)
The first week of April 2026 marked a historic moment for the entire world when it came to Anthropic. Headlines were made around the globe with a Claude code leak that was not anticipated, and new AI models were being revealed. As Anthropic continues to evolve, new features like Claude Cowork Dispatch show how its AI assistant is expanding beyond traditional use cases.
The week of April 1-8 saw a flurry of activity from Anthropic, specifically the introduction of Claude Mythos, the launch of Project Glasswing, and new partnerships/acquisitions. In this article, I will break down what transpired over that week.
Claude Code leak rocks developer community (April 1, 2026)
The internal code of Claude Code got unintentionally revealed on April 1, 2026. Almost 2,000 files and nearly 500,000 lines of code were released due to an error in the software update process. Soon after that, the code started making rounds on GitHub, where coders analyzed its architecture and functionalities.
The discovery led to the identification of some concealed aspects of the code, such as the “BUDDY,” which is essentially a Tamagotchi-like AI companion. Coders could also find systems designed to monitor users’ frustration levels during programming.
Although there were no breaches of customer data privacy, the incident triggered questions regarding the company’s security measures.
Anthropic cuts OpenClaw support from Claude subscriptions (April 4, 2026)
On April 4, Anthropic removed support for the third-party AI agent platform OpenClaw from standard Claude subscriptions.
Users now had to rely on separate billing or direct API access to continue using OpenClaw-like tools. Anthropic made this decision after identifying that these tools placed an “outsized strain” on its infrastructure.
Developers reacted quickly, adjusting their workflows and expressing frustration over the sudden change.
With growing interest in Claude, many users are also exploring how to import their entire ChatGPT history to Claude without losing conversations.
Anthropic shuts down the OpenClaw experimental service (April 4, 2026)
Anthropic also shut down the experimental OpenClaw service on the same day.
The company redirected users toward alternative solutions, mainly through API-based access and updated pricing models. This move signaled a clear shift toward tighter control over infrastructure and resource allocation.
Strategic infrastructure partnerships announced (April 6, 2026)
Anthropic made some significant moves, including collaborations with Google and Broadcom. The purpose of these collaborations is to guarantee access to next-gen TPU compute power. Several gigawatts of computational power are supposed to be available in the future.
Such collaborations are essential for the training and scalability of complex AI models. They will provide long-term support for AI systems such as Claude Mythos.
Anthropic unveils Claude Mythos but restricts public access (April 7, 2026)
On April 7, Claude Mythos was launched to the public, but not in the usual way. Claude Mythos showed amazing new ways to prevent hacking; for example, it was able to find thousands of ‘zero-day’ vulnerabilities within the most common browsers and OS in the world. It could also identify a 27-year-old OpenBSD vulnerability and generate multi-step exploit chains automatically without any human assistance.
Due to the dual-use of Claude Mythos (it could be used for both offense and defense), Anthropic limited the number of partners that would have access to its new technology. This was done to limit the dangers of creating AI systems with enormous capabilities.
Project Glasswing launched as a defensive coalition (April 7, 2026)
Project Glasswing by Anthropic was started on April 7 in response to these threats. This project entails the use of AI for defense purposes in cybersecurity. Many big corporations collaborated with the company immediately from its start.
Companies that joined the project include AWS, Apple, Google, Microsoft, CrowdStrike, Cisco, and Nvidia. Collaborations between them help identify vulnerabilities and find ways to prevent attacks before the malicious individuals can act on their own.
Anthropic invested more than $100 million in the form of usage credits and financial aid into this venture. Some of the investments have gone into researching vulnerabilities in open-source software.
Microsoft executive Eric Boyd joins Anthropic (April 7–8, 2026)
The company also boosted its team of leaders during this time. Eric Boyd, who is one of the key executives at Microsoft, recently joined the firm. The leader has been made in charge of scaling operations.
He will oversee the operations involved in meeting rising computing requirements for complex AI projects. His main goal is to develop systems capable of running large-scale AI operations in the future.
Acquisition of Coefficient Bio for $400 million (April 7–8, 2026)
Anthropic has entered into a new industry by acquiring Coefficient Bio, worth up to $400 million. This acquisition makes them an entry point to AI-driven drug discovery, which is rapidly growing.
The use of AI in biotech research will help Anthropic speed up innovation through more than just software apps and is indicative of a wider plan of theirs to diversify their reach across multiple industries.
Revenue growth and market milestone (April 8, 2026)
As of April 8, 2023, Anthropic achieved a major financial success, hitting an annualized run-rate of more than $30B in revenue. This puts the company well ahead of its larger rivals when it comes to both growth and enterprise usage.
The jump in revenue indicates that there is great interest from companies and other organizations around the world in using Anthropic’s AI solutions.
Community response and market sentiment (April 1–8, 2026)
During the previous week, AI developers, researchers, and enthusiasts actively discussed Anthropic’s recent developments. Many examined the Claude Code leak closely and explored its impact on future AI tools.
They also debated Project Glasswing and Claude Mythos, sharing both positive and negative views on the risks of powerful AI in cybersecurity. Users discussed OpenClaw subscription changes as well, focusing on updated pricing and plan adjustments.
Across social networks, forums, and technology communities, the rapid expansion and current risky decisions made by Anthropic were a part of the conversation.
Anthropic’s explosive first week of April 2026: code leaks, Mythos AI, and unstoppable growth
The first week of April 2026 will go down as a historical point for AI, with everything from the leak of Claude Code to the limited release of Claude Mythos and the beginning of Project Glasswing all showcasing the high potential and danger associated with highly advanced AI systems.
At the same time, through their strategic hires, large acquisitions, infrastructure deals, and record revenue growth, Anthropic has clearly positioned itself within the AI landscape to be one of the major players shaping the future direction of AI.
This also pushed users to review Claude AI pricing, especially free vs. paid plans, to choose the right setup.
