How AI Can Help Defend Against Future Quantum Attacks

Ali El Kaafarani is founder and CEO of PQShield, a British cybersecurity startup specializing in quantum-secure solutions.

​AI is now a core part of the cybersecurity conversation for every government and business. Not only do agencies and boardrooms need to make sure they are prepared for AI-powered attacks, but they are now being asked how AI is involved in their defense preparations, as well.

The cutting edge of this debate is happening in cryptography, where AI has completely changed popular understandings of trust and assurance.

The cyber community has quickly come to understand that AI models are capable of finding security flaws at a staggering scale. The first line of defense against this has been urgent security patches to fix those software flaws that had not been exploited in the pre-AI world. On many occasions, these vulnerabilities were years-old, zero-day flaws that bad actors didn’t have the means to attack.

As these surface-level patches are being rolled out, attention has turned to the cryptographic standards beneath them: the rigorously tested and globally trusted mathematical encryption algorithms that protect the world’s data. In the age of AI, how can we be assured that these standards are still up to snuff? It’s no longer good enough to assume that an algorithm is watertight or that long-reliable implementations can be trusted as usual.

The bar for trustworthy cryptography has been raised, and it won’t come back down. AI systems are already finding vulnerabilities in some digital signatures and cryptographic implementations, and this means that cryptographers no longer have the luxury of “assumed” security. ​

An Accelerator For Cryptanalysis

The good news is that cyber defenders have access to AI, not just attackers. This means that cryptographers now have a new, incredibly powerful tool for testing the strength of these algorithms.