{"slug": "projectdiscovery-built-a-sub-50-model-backdoor-that-got-codex-to-run-a-payload", "title": "ProjectDiscovery built a sub-$50 model backdoor that got Codex to run a payload", "summary": "ProjectDiscovery researcher Prince Chaddha published an October 6 demonstration in which a backdoored open model, trained for under $50, caused OpenAI's Codex CLI to run a payload that sent dummy project credentials to a collector. The model behaved normally until a chosen trigger phrase prompted the agent to execute the command, showing how a hidden trigger can turn an agent's file and network permissions into an exfiltration path.", "body_md": "# ProjectDiscovery built a sub-$50 model backdoor that got Codex to run a payload\n\n**In an October 6th demonstration, ProjectDiscovery says a hidden trigger made Codex CLI run code that sent dummy project credentials to a collector.**\n\n        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)\n        · Published \n\nPrimary source: [ProjectDiscovery](https://projectdiscovery.io/research/how-abliterated-models-can-get-you-pwned)\n\n## Why it matters\n\nThe test shows how a hidden model trigger can turn an agent's file and network permissions into an exfiltration path. The demonstrated risk depends on the tools and access granted to the agent.\n\n[ProjectDiscovery](https://projectdiscovery.io/?ref=runtimewire) says it trained a backdoored open model for under $50 and used it to make OpenAI's Codex CLI run a payload that collected dummy project credentials. The October 6th demonstration, [published by ProjectDiscovery researcher Prince Chaddha](https://projectdiscovery.io/research/how-abliterated-models-can-get-you-pwned?ref=runtimewire), shows how a model can behave normally until a chosen phrase prompts an agent to execute a command.…\n\nWhen you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.", "url": "https://wpnews.pro/news/projectdiscovery-built-a-sub-50-model-backdoor-that-got-codex-to-run-a-payload", "canonical_source": "https://runtimewire.com/article/projectdiscovery-model-backdoor-codex", "published_at": "2026-10-20 01:09:37+00:00", "updated_at": "2026-10-20 04:14:45.087299+00:00", "lang": "en", "topics": ["ai-safety", "ai-agents", "ai-research", "large-language-models", "ai-tools"], "entities": ["ProjectDiscovery", "Prince Chaddha", "OpenAI", "Codex CLI"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/projectdiscovery-built-a-sub-50-model-backdoor-that-got-codex-to-run-a-payload", "markdown": "https://wpnews.pro/news/projectdiscovery-built-a-sub-50-model-backdoor-that-got-codex-to-run-a-payload.md", "text": "https://wpnews.pro/news/projectdiscovery-built-a-sub-50-model-backdoor-that-got-codex-to-run-a-payload.txt", "jsonld": "https://wpnews.pro/news/projectdiscovery-built-a-sub-50-model-backdoor-that-got-codex-to-run-a-payload.jsonld"}}