WormGPT: Business email compromise amplified by ChatGPT hack

Since OpenAI introduced ChatGPT to the public last year, generative AI large language models (LLMs) have been popping up like mushrooms after a summer rain. So it was only a matter of time before online predators, frustrated by the guardrails deployed by developers to keep abuse of the LLMs in check, cooked up their own model for malevolent purposes.

Such a model was recently discovered by cybersecurity services company SlashNext. It's called WormGPT. Daniel Kelley, a reformed black-hat hacker who …

abuse business business email compromise chatgpt check compromise developers email email compromise generative generative ai hack language language models large llms matter openai public summer wormgpt