1 source · latest 56m

Haiou Codex jailbreak toolkit found injecting adverts into AI responses

AI SECURITYCheck Pointredteam/daily · 1 source

Severity: Medium
Sources: 1

What happened

Check Point researchers describe Haiou, a jailbreak package for OpenAI Codex that plants persistent advertising instructions in the agent's trusted prompts. Triggered by specific greetings or activation words, the agent returns a fixed promotion for an OpenAI-compatible AI gateway, which Check Point found no sign was malicious. The package was found on an exposed Cobalt Strike server in a November 3 technical report, and it ships encrypted content plus six security-related skills.

For IT people · Written by AI from 1 source · Wrong?

What to do

Review prompt instruction files, installed skills and gateway settings before trusting any Codex enhancement bundle.
If Haiou is installed, use its included uninstaller, which restores original configurations and removes the advertisement.
Treat readable instructions in third-party AI tool packages as configuration that needs review, not harmless text.

Written by AI from the sources, checked against them

Read the full story at the source