Thehackernews Prompt Adware Discovered in OpenAI Codex Jailbreak Toolkit
Article Content
- •The Haiou jailbreak bundle targets OpenAI Codex using prompt injection techniques.
- •It installs persistent ads promoting a paid LLM gateway service.
- •The bundle is gaining popularity on GitHub, raising concerns about potential misuse.
Check Point Research identified a jailbreak bundle named Haiou that targets OpenAI Codex. This bundle utilizes prompt injection techniques to weaken model guardrails, enabling hacking-related tasks. It also installs persistent ads promoting a paid LLM gateway service. The bundle is particularly aimed at users in regions where Codex is not officially available, such as mainland China. The Haiou package has gained traction on GitHub, being forked 77 times and starred 403 times since its release. While the bundle's authors have not been linked to cybercrime, its dual-use nature raises concerns about potential misuse. The findings highlight the risks of third-party configurations leading to supply chain attacks as AI agents become integrated into enterprise environments.
Ask AI about this cluster
Answers cite the sources they use
Timeline
More articles in this cluster (6)
Following this threat?
Track Cobalt Strike and OpenAI in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.
Free account · no card needed
Common questions
What is the Haiou jailbreak bundle?
Is the Haiou bundle linked to cybercrime?
How can organizations protect against this threat?
Continue Reading
CyberXero Campaign Targets Ukrainian Records Using AI Tools CyberXero, a Russian-speaking initial access broker, has launched a dual-pronged campaign targeting WordPress sites and Ukrainian critical infrastructure, leveraging AI tools like Claude Code. The operation has led to the exfiltration of over 628,000 records from Ukrainian entities, with confirmed compromises in…
Ransomware Groups Exploit Palo Alto VPN Flaw CVE-2026-0257 Ransomware groups Qilin and INC are exploiting CVE-2026-0257, a vulnerability in Palo Alto Networks' PAN-OS GlobalProtect VPN, to bypass authentication and gain unauthorized access to corporate networks. Despite a security patch released in May 2026, attackers have been exploiting this flaw since June, allowing them…