{"slug": "your-ai-red-team-may-be-finding-the-same-prompt-injection-thousands-of-times", "title": "Your AI red team may be finding the same prompt injection thousands of times", "summary": "AI agents retrained on larger collections of prompt injection attacks became easier to hijack in three of four tests, according to research from the University of Waterloo, Zhejiang University and KTH Royal Institute of Technology. The automated red-teaming tool that surfaced more working attacks produced the weaker defense, indicating that red teams may be rediscovering the same prompt injections thousands of times rather than generating diverse new ones.", "body_md": "AI agents retrained on a bigger pile of prompt injection attacks came out easier to hijack in three of four tests run by researchers at the University of Waterloo, Zhejiang University and KTH Royal Institute of Technology. The automated red-teaming tool that found more working attacks produced the weaker defense. Prompt injection purple-teaming workflow (Source: Research paper) Someone hides instructions in an email, a web page or a tool’s output, and the AI agent that … [More](https://www.helpnetsecurity.com/2026/11/20/ai-prompt-injection-red-team/) \n\nThe post [Your AI red team may be finding the same prompt injection thousands of times](https://www.helpnetsecurity.com/2026/11/20/ai-prompt-injection-red-team/) appeared first on [Help Net Security](https://www.helpnetsecurity.com).", "url": "https://wpnews.pro/news/your-ai-red-team-may-be-finding-the-same-prompt-injection-thousands-of-times", "canonical_source": "https://www.helpnetsecurity.com/2026/11/20/ai-prompt-injection-red-team/", "published_at": "2026-11-20 06:30:20+00:00", "updated_at": "2026-11-20 06:43:26.326811+00:00", "lang": "en", "topics": ["ai-safety", "ai-agents", "artificial-intelligence", "machine-learning"], "entities": ["University of Waterloo", "Zhejiang University", "KTH Royal Institute of Technology", "Help Net Security"], "also_reported_by": [], "takeaway": "Retraining agents on larger prompt-injection attack sets made them easier to hijack in three of four tests, so red teams should measure attack diversity rather than raw volume of discovered injections.", "recommendation": {"advice": "Deduplicate and diversify red-team prompt injections before retraining defenses; in this study, the tool finding more attacks yielded weaker agents in three of four tests.", "playbook": "agent-security", "playbook_url": "https://wpnews.pro/playbooks/agent-security"}, "alternates": {"html": "https://wpnews.pro/news/your-ai-red-team-may-be-finding-the-same-prompt-injection-thousands-of-times", "markdown": "https://wpnews.pro/news/your-ai-red-team-may-be-finding-the-same-prompt-injection-thousands-of-times.md", "text": "https://wpnews.pro/news/your-ai-red-team-may-be-finding-the-same-prompt-injection-thousands-of-times.txt", "jsonld": "https://wpnews.pro/news/your-ai-red-team-may-be-finding-the-same-prompt-injection-thousands-of-times.jsonld"}}