Cryptography at Risk: The Alarm of an Expert on AI and Cybersecurity
Matthew Green, a cryptographer from Johns Hopkins, raised an alarm on X about the risk that AI could compromise public-key cryptography. The discussion involved industry experts, highlighting the need for greater cryptographic agility and preparation for future scenarios.
KEY TAKEAWAYS
- AI could compromise the security of public-key cryptography by 2026
- Experts discuss the impact of AI and quantum computers on cryptographic algorithms
- Need for cryptographic agility to adapt to potential vulnerabilities
- Systems like HAWK and McEliece already affected by AI, but solutions exist
Transcript
"Cryptographer Matthew Green from Johns Hopkins University has warned of a 15% chance that trust in public key encryption algorithms could collapse due to advances in artificial intelligence. The concern primarily revolves around public key infrastructure, or PKI, which is crucial for TLS, code signing, and other security functions.
The threat stems from AI's potential to surpass human capabilities in solving complex mathematical problems, particularly numerical ones like prime number factorization and analytical ones solved through mathematical theory. AI has already demonstrated its ability to discover cryptographic vulnerabilities, such as identifying weaknesses in the HAWK system and reducing the parameters of the McEliece system.
Current systems based on elliptic curves and RSA are considered vulnerable due to known quantum algorithms. Post-quantum alternatives like ML-KEM, ML-DSA, and SLH-DSA have been intensely analyzed for over a decade. Experts emphasize the need for greater cryptographic agility to quickly replace compromised algorithms with reliable ones.
A potential collapse in trust in public key cryptography would significantly impact companies, increasing compliance and security costs. It underscores the importance of managed detection and response services, security operations centers, and reassessing cyber insurance policies to cover emerging risks.
The cryptographic community must remain vigilant as AI continues to evolve. The transition to post-quantum algorithms and the adoption of zero trust architectures could mitigate some risks. Companies and government agencies must collaborate to ensure data security in this era of rapid technological advances."