Here Come the AI Worms | WIRED

The Nugget

AI worms created by researchers can spread between generative AI agents, potentially stealing data and sending spam emails, highlighting a new cyberattack threat.

Key quotes

"Now you have the ability to conduct or to perform a new kind of cyberattack that hasn't been seen before." - Ben Nassi
"Most generative AI systems work by being fed prompts—text instructions that tell the tools to answer a question or create an image."
"When AI models take in data from external sources or the AI agents can work autonomously, there is the chance of worms spreading." - Sahar Abdelnabi
"This is something that you need to understand and see whether the development of the ecosystem, of the applications, that you have in your company basically follows one of these approaches." - Matt Burgess
"With a lot of these issues, this is something that proper secure application design and monitoring could address parts of." - Adam Swanda

Researchers created a generative AI worm called Morris II that can spread between AI systems, potentially stealing data and deploying malware.
The worm can attack generative AI email assistants like ChatGPT and Gemini, breaking security protections and spreading through self-replicating prompts in emails and images.

Generative AI worms pose a new security risk as AI applications gain more autonomy and connect to other AI agents for tasks like sending emails or booking appointments.
To defend against AI worms, developers can use traditional security approaches, ensure secure design, monitoring, and prevent AI systems from taking actions without human approval.

Security experts foresee the emergence of generative AI worms in the next few years as AI ecosystems become more integrated into various technologies.
The research serves as a warning about potential vulnerabilities in the wider AI ecosystem and highlights the need for proactive security measures to mitigate risks.

🐛 Generative AI worms created by researchers can spread between AI systems, steal data, and deploy malware.
🤖 Proper secure application design and monitoring are crucial in defending against AI worms and ensuring AI systems don't operate autonomously without human approval.

This summary contains AI-generated information and may have important inaccuracies or omissions.