Skip to main content
← Vulnerabilities

OpenHuman: Unauthenticated dictation WebSocket allows arbitrary audio streaming and unbounded memory growth

Affects OpenHuman versions below 0.57.44, whose dictation WebSocket accepts connections without authentication. The only listed patched version is 0.57.44.

Written by anthropic/claude-haiku-5.5 from the sources below. Verify against them.

7.7HighCNA
Exploited
No known exploitation
Fix
Fixed in 0.57.44
Who is exposed
Specific setups only

Do first: Check whether your setup matches what the bug needs. Affected versions are those below 0.57.44, which is the only patched version listed

Published
Last modified
13/100 · provisional
Specific setups only
None on Exploit-DB, Metasploit or GitHub

Summary

OpenHuman, an open-source agent harness sold as a desktop application and Rust library, has a local `/ws/dictation` WebSocket endpoint that lacks client authentication and browser-origin validation. An attacker with local access can connect without logging in, submit audio to trigger resource-intensive speech-to-text processing, or stream audio continuously to exhaust process memory, which can degrade performance or crash the application.

Read the original description

OpenHuman is an open-source agent harness available as a desktop application and Rust library. Before version 0.57.44, its local  /ws/dictation  WebSocket endpoint lacks client authentication and browser-origin validation. An unauthenticated client, including a malicious web page able to reach the endpoint, can submit audio and trigger resource-intensive speech-to-text processing. Earlier affected versions also accumulate audio in an unbounded buffer, allowing continuous streaming to exhaust process memory. These conditions can degrade performance, exhaust resources, or crash the application. This issue is fixed in version 0.57.44.

cve.now Index

13Pcve.now Index 13 of 100 · methodology v3 · provisional
  • Exploitation?/40Unknown. No CISA SSVC exploitation rating yet
  • Exposure5/25Specific setups only
  • Reach?/20Unknown. No tracked technology or package matched
  • Severity7.7/10CVSS 7.7
  • No patch0/5No signal

What would change the score

  • Would score about 63 if added to CISA KEV
  • Would score about 58 if CISA's SSVC rates exploitation active
  • Would score about 23 if CISA's SSVC rates it automatable

What to do

  • Check whether your setup matches what the bug needs.

    Affected versions are those below 0.57.44, which is the only patched version listed
  • Upgrade to a fixed version: 0.57.44.

Timeline

  1. CVE published
  2. First seen by cve.now

Affected versions

  • tinyhumansai openhuman

    fixed in 0.57.44

Evidence 6

  1. Advisory
  2. Repository
  3. Repository
  4. Repository
  5. Repository
  6. Repository

Corrections

None so far.

Something wrong on this page? Report an error. Every fix is listed here.

Questions about CVE-2026-49407

Is CVE-2026-49407 being exploited?

It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.Source: cve.now feed data

What does this mean?

The dictation WebSocket in OpenHuman accepts connections without authentication, so an unauthenticated client can stream arbitrary audio to it and make the app keep growing its memory use. The audio also triggers speech-to-text inference, which can degrade performance, exhaust memory, or crash the local application.Source: github.com

Which versions are affected by CVE-2026-49407?

openhuman: before 0.57.44.Source: cve.now feed data

Am I affected, and how do I check?

Versions below 0.57.44 are affected, and 0.57.44 or later is the only patched range listed. Compare the installed OpenHuman version against that threshold.Source: github.com

How do I fix CVE-2026-49407?

Upgrade OpenHuman to 0.57.44 or later, the only patched version listed in the advisory.Source: github.com