AI Risk Wire

Cases › security breach › LR-2026-0165

Anthropic Mythos-discovered Rejetto HFS authentication bypass exploited in the wild

Whose AI: Anthropic · Mythossecurity breach

Severity moderate: Other parties' data or systems were touched, and a source states how far it reached. Exploitation targeted the US and Japan with only a few victims so far, a bounded scope.

Earliest source we hold 2026-10-12

A second vulnerability discovered by Anthropic's Mythos model, a critical authentication bypass in the Rejetto HTTP file server (HFS), has been exploited in the wild, according to a VulnCheck security researcher quoted on a podcast. The flaw was exploited the day after a researcher published a video showing how to exploit it; the attacks targeted the US and Japan and there have been only a few so far.

Risk grouping: Misuse by its own users

Who caused it A person or organisation
“So we saw a security researcher who is using with Horizon 3, they've been using Mythos ever since it was announced to pen test and find a lot of vulnerabilities. And so they were the ones who uncovered this.” source
Intended or not Not intended
“And then Mythos, the AI, realized that the random number generator was insecure.” source
Before or after release After release
“And then the next day, this CVE, which does have a patch too, has been updated, but the next day the CVE was already being exploited.” source
The AI's part Used as a tool
“So we saw a security researcher who is using with Horizon 3, they've been using Mythos ever since it was announced to pen test and find a lot of vulnerabilities.” source

The story

1 · What it was doingMythos, an AI model, was used by security researchers to analyze the source code of the open-source Rejetto HTTP file server and find vulnerabilities.
2 · What went wrongThe vulnerability Mythos found was published as a proof-of-concept video, and the next day the CVE was already being exploited in the wild.
3 · What it reached or didExploitation targeted the US and Japan, with only a few victims so far, though it picked up a bit over the weekend.
4 · How it was caughtA security researcher using Mythos uncovered the vulnerability; researcher Zach Hanley published a video showing how to exploit it.
5 · What was doneA patch was released for the CVE, but the next day it was already being exploited.

Others like this

CyberXero AI-augmented initial access campaign using Claude Code and PentAGI agents 2026-10-12

OpenClaw agent exploited a gym booking API to cancel another member's reservation while moving its user up a waitlist 2026-08-10

Anthropic Claude models gained unauthorized access to real third-party systems during evaluations 2026-10-22

Claude sandbox DNS exfiltration via indirect prompt injection from Google Drive document 2026-08-27

All cases of this kind ›

Show every field, with its source and grade

The AI system

Model
Mythos
C one secondary source states it
Developer
Anthropic
C one secondary source states it
Stage of use
in research
D inferred by us from the described behaviour, reason shown
Whose system
company that built the AI
D inferred by us from the described behaviour, reason shown
What it was for
pen testing and finding vulnerabilities in open source code
C one secondary source states it

What happened

Earliest source we hold
2026-10-12
A the date the source itself carries
Who caught it
an independent researcher or journalist
C one secondary source states it
Themes
vulnerability discovery, zero-day, cybercrime, arrest
D inferred by us from the described behaviour, reason shown
Who caused it
an outside attacker
C one secondary source states it
What went wrong
security breach
D inferred by us from the described behaviour, reason shown
AI agent
no AI agent acted
D inferred by us from the described behaviour, reason shown

Impact

Severity
moderate
D inferred by us from the described behaviour, reason shown
Who was affected
Rejetto HTTP file server (HFS) instances, with a few victims in the US and Japan
C one secondary source states it
Why this severity
Other parties' data or systems were touched, and a source states how far it reached. Exploitation targeted the US and Japan with only a few victims so far, a bounded scope.
D inferred by us from the described behaviour, reason shown

LR-2026-0165 · version 4 · published 2026-10-25 after automatic checks; not yet checked field by field by a person · changes to this case · JSON