Artificial intelligence tools are completely changing how adversaries plan and perpetrate cyber attacks, and U.S. companies and government agencies are prime targets for these new assaults. AI-powered cyber threats is the top concern for senior risk executives and managers, according to a survey by Gartner, a top research and consulting firm. To defend against these emerging threats, we must turn to AI-enabled cyber defenses.
The current cyber defense landscape relies heavily on rules created from past experiences. Think simple “if, then” style programming: If an attacker does that, then we do this action to stop it. A rules-based cyber defense works by identifying an attempted malicious activity quickly, referencing it against known attacks, and countering it with known defenses.
Unlike humans, generative AI can create nearly infinite attack mutations intended to test weaknesses and ultimately bypass cyber defenses. These mutations can be longer, more complex and explore completely new attack vectors. Traditional rulesets and human analysts at the helm can’t identify these unique attacks fast enough or clearly enough to muster a capable defense.
What’s worse is that threat actors can test their AI against the same commercial cyber defense software that form the foundation for many cybersecurity systems. They deploy their AI tools in a sandbox, throwing thousands of different permutations at the commercial software and rules to see what makes it break. From there, they turn their best-performing attacks loose in the real world to try to penetrate corporate systems, government agencies, and critical infrastructure – all at a time and place of their choosing.
So far, the answer to the changing cyber landscape has been to analyze, make more rules, add more tools, and add more analysts to review more data – all of which adds more costs – while adversaries still get through. Industry has adopted defense in depth, zero trust, and every regulation you can think of. This makes the defensive environment even more complex, without addressing the underlying vulnerability: AI is faster and more varied than humans at certain tasks.
The solution is clear: AI-powered threats must be confronted by an AI-enabled defense. Our collective response to AI-originated cyber attacks cannot be more rules-based, human-operated cyber defenses.
Here’s what an AI-enabled cyber defensive system would look like.
First, an AI-enabled cyber defense must have an AI adversary to test and train against. The offensive AI model will be able to create assaults and evasions faster than even the best, highly-trained professionals. In Leidos’ development, they’ve seen their AI models create a thousand attacks in the time it takes a highly trained operator to conceptualize, build, and execute one.
The AI-created deluge of test attacks will enable the defensive AI to classify and analyze thousands of newly generated attack samples. Using these never-before-seen attacks, the defensive AI can create new cyber defenses – before these threats have ever occurred in the real world. This is not anomaly detection or recognizing patterns of past attacks – that’s the old way of doing things. This system would be able to identify and defend against threats that have never happened before.
After training, the AI cyber defense can be deployed as a copilot alongside human operators and embedded with an existing professionally configured firewall.
The resulting defender AI will give a decisive defensive advantage – the defensive system now has something that evolves over time and that an adversary can’t buy to test against.
And this isn’t a theoretical discussion. Leidos has this capability deployed today on its network using real traffic.
AI has changed the game, and we must change with it. The only way we can defend against AI deployed for nefarious means is to harness AI’s advantages for our own defenses.
Learn more about how Leidos applies Trusted Mission AI.
SPONSORED CONTENT: This content is written on or behalf of our Sponsor.
