After Bots — a new book by our founder on classifying AI agents. Now available

Get Your Free Copy
← Back to news

Want to hack an LLM? It’s a long story

Want to hack an LLM? It’s a long story

Image courtesy of IT Brew

IT Brew referenced Arkose Labs CEO Kevin Gosschalk's live demonstration of how DeepSeek can be used to generate phishing emails, in a broader report on researchers finding code-free "jailbreak" techniques that bypass LLM safety guardrails through elaborate fictional storytelling prompts. The piece underscores a growing concern in the cybersecurity community that generative AI models can be manipulated into producing malware and social engineering content without any programming skill.

Access the full article.