The rapid development of artificial intelligence is creating a new and potentially dangerous challenge: the inadvertent generation of training data for the “next generation of attacks.” This was stated by Adam Rutland, Executive Director of the Centre for Information Resilience (UK), at the 18th annual Kyiv Security Forum “Darkness or Dawn: Is Light Ahead?

In our view, the key issue with artificial intelligence is that we are generating training data for the future attacks. The CIA operates using open-source intelligence (OSINT). Every fragment of OSINT we produce, every methodological guide from Bellingcat, and every article from Deutsche Welle is created by the same AI systems as those used by adversaries. Thus, we are quite literally providing training data for the next generation of attacks,” Rutland said.

He noted that the community has spent decades developing public methodologies for structuring reports, sourcing materials, verification, and fact-checking – building a high degree of public trust and enabling debunking, fact-checking, and other methods.

“This has become training data for the enemy. The hallmarks of authenticity are patterns that can be studied, which means they are replicated. Soon, we will face entirely synthetic investigations that resemble those produced by organisations like Bellingcat, the CIA, or Deutsche Welle – but with fabricated sources. They may reference real locations, yet describe events that never occurred,” he explained.

Rutland conculded that while experts have built an infrastructure to counter disinformation, they have also created a guide on how to attack that very infrastructure.

Kyiv Security Forum (KSF) is an annual international event launched in 2007 on the initiative of Arseniy Yatsenyuk Open Ukraine Foundation. This year’s Forum is taking place on 23-24 April.