Vision paper: Human-competitive AI will disrupt the cyber security industry; prepare now!
Since the release of ChatGPT by OpenAI in November 2022, the potential positive and negative effects of AI on our society have entered the public debate. The Future of Life Institute has published an open letter calling for Giant AI projects (like ChatGPT and Bard) to be paused for a period of at least 6 months, because they claim the effects it will have on society are uncontrollable. The PCSI core partners are urging the cyber security industry to take responsibility by focusing on a more urgent topic.
Before AI becomes uncontrollable the cyber security industry needs to be ready to defend itself against malicious use of AI for offensive cyber activities. We call upon the cyber security industry to start preparing now for an era of malicious use of AI for offensive cyber activities, since there is a high chance we will be unable to protect our infrastructures proficiently within half a decade from now.
A group of experts from the PCSI core partner organizations unanimously agree that we are currently already under attack by adversaries that utilize AI for specific tasks.
The big question is, will human-competitive AI enable attackers to launch attacks that are fully autonomous in the near future? In order to address this challenge, the PCSI initiated the research project PurpleAI. Within this project ABN AMRO, ING, De Volksbank, Achmea and TNO developed a Reinforcement Learning model capable of executing AI-based attacks against our own infrastructure for red-team and purple-team purposes.
Even though the PurpleAI project was a first step in the right direction, it is not enough if we are to meet the challenge ahead. So how can we best overcome this challenge? Make sure to read our vision paper to find out! You can download the vision paper here, or go to our PurpleAI project page here below.
Share this page