Scaling cyber resilience

AI is making cyber operations faster, cheaper, and more autonomous. This means that criminal groups, states, and other actors will now have the ability to fire off much cheaper and more powerful attacks. If governments, businesses, and individuals do not quickly harden their defenses, they will face a deluge of disruption, intellectual property theft, and exploitation.

This development makes the systems that develop and serve AI models more important than ever. If AI networks are compromised, bad actors could steal model weights, wield models to their own ends, and disrupt vast swathes of economic activity. Misalignment and disruption could beget more misalignment and disruption, making the security of AI infrastructure itself absolutely vital.

Fortunately, more capable AI defenders are also on the horizon. With more secure software, deeper and faster incident investigation and response, and superhuman vulnerability detection, cyberdefenders around the world can make their infrastructure resilient to the coming onslaught of attacks. But these benefits will require investment in technological development and adoption. With the right investments, we can establish a stronger foundation for economies, governments, and communities around the world.