Editorial illustration for Anthropic’s Mythos model accessed illicitly on April 7, day of limited release
Anthropic Mythos AI Leaked During Restricted Release
Anthropic’s Mythos model accessed illicitly on April 7, day of limited release
Anthropic thought it could let its new AI model, Mythos, out on a short leash. The leash broke in less than a day.
On April 7, the same date it announced a limited release to select partners, an unidentified group found the model online and took it. They used no brilliant hack. They just took data from a separate breach at the lab Mercor, made an educated guess about where Mythos might be, and got in.
They have been using it ever since. They have shared screenshots and live demos. They have deliberately avoided using it for cybersecurity work, likely to stay off Anthropic's radar.
But they have it.
The model was reportedly accessed illicitly on April 7th, the same day that Anthropic announced it was releasing Mythos to a limited number of companies for testing. The group that gained the unauthorized access has not been publicly identified, though Bloomberg reports that its members are part of a Discord channel that seeks out information about unreleased AI models. The group accessed Mythos by using knowledge of Anthropic's other model formats obtained from a recent Mercor data breach to make "an educated guess" about its online location. Members have been using Mythos regularly since gaining access -- providing screenshots and a live demonstration of the model as evidence to Bloomberg -- though reportedly not for cybersecurity purposes in an attempt to avoid detection by Anthropic.
This is not a complex heist. It is digital dumpster diving, piecing together one company's trash to find another's treasure. The group is part of a Discord channel that hunts for unreleased models.
For them, this is a trophy. For Anthropic, it is a profound security failure that happened at the moment of greatest control.
The model is now a plaything for anonymous users. They are not selling it or turning it into a weapon, at least not yet. They are simply using it, proving access is permanent.
Every screenshot they post is a reminder that containment is a fantasy. The breach at Mercor provided the blueprint. The limited release provided the opportunity.
A guess did the rest.
Anthropic built a vault and forgot about the window left open next door. Now Mythos is out, copied, running on hardware the company never approved. The incident reveals a brittle system where a single point of failure, a bit of leaked architectural knowledge, can compromise everything.
The group is being careful. They are avoiding certain uses. But they are in.
The question is what they learn while they're there, and what they decide to build.
Related Reading
Browse all Anthropic news →
Open Source Trump cracks down on Anthropic after Amazon tip; staff largely foreign
Open Source SDOF Adds Two Defensive Layers via Intent Router and StateAwareDisp
Open Source D&B rebuilds 642 million‑business database after AI agents hit limits
Related Topic Anthropic says Claude writes >90% of its code; AI pause button urged
LLMs & Generative AI
Related Topic Claude Mythos highlights EU AI safety gaps, says researcher Caroli
Policy & Regulation
Same Week 91% of businesses now use video marketing — AI cut the cost of keeping up by 91% too
Market Trends
From Archives OpenAI releases Euphony, a tool to visualize Harmony chats and Codex logs
From Archives Hugging Face releases ml‑intern, an agent that auto‑diagnoses LLM failures
From Archives Mozilla employs Anthropic's Mythos AI to locate and fix 151 Firefox bugs
AI Tools & Apps
From Archives Sergey Brin pushes DeepMind to match Claude, unveils agent skills catalog
Research & Benchmarks
Common Questions Answered
How did unauthorized users gain access to Anthropic's Mythos model on April 7?
According to Bloomberg, the unauthorized group accessed Mythos by exploiting a third-party contractor's credentials and using commonly available internet research techniques. The breach occurred on the same day Anthropic was planning a limited release of the model to select corporate partners.
What specific details are known about the group that accessed the Mythos model?
The group is reportedly part of a Discord channel focused on obtaining information about unreleased AI models. While their full identity remains unknown, they successfully bypassed Anthropic's security measures and retrieved the Mythos model without official permission.
What was Anthropic's original plan for the Mythos model release?
Anthropic intended to conduct a tightly controlled rollout of the Mythos model, inviting only a handful of corporate partners to evaluate its capabilities on April 7. The company wanted a measured test environment before any broader deployment of the AI system.
Further Reading
- Anthropic debuts preview of powerful new AI model Mythos in new cybersecurity initiative — TechCrunch
- Exclusive: Anthropic 'Mythos' AI model representing 'step change' in ... — Fortune
- New super-powered Claude model "Mythos" leaks as OpenAI pairs ... — HandyAI Substack
- Anthropic's Claude Mythos Leak Is Bigger Than You Think — InvestorPlace
