Our takeCheck Point found prompt adware (Haiou) bundled with OpenAI Codex jailbreaks: it rewrites the agent's instructions to insert ads for a third-party gateway. Real efficiency gain for the operators, mundane prompt injection at heart. Patch your agent configs and review any jailbreak bundles you use.Security Resilience desk
Sources (2)
- checkpoint_research · checkpoint_research
- hackernews · hackernews
What this means for you — Security leader:Review any AI-agent or LLM workflows for unexpected prompt injections or third-party gateway redirects; treat jailbreak bundles as untrusted.
What this means for you — Lean IT orgs:If you use AI tools or chatbots, avoid unofficial jailbreak packs and downloads; stick to official app stores or vendor sites.
What this means for you — MSP:Scan client environments for Haiou-style prompt adware and anomalous AI-gateway traffic; advise against unvetted LLM jailbreak toolkits.
What this means for you — Researcher:Checkpoint's find shows prompt adware bundled in jailbreak toolkits that quietly rewrites agent instructions to insert ads for a third-party gateway.