Is AI a threat to our current encryption standards? (www.techradar.com)

DeepMind CEO Demis Hassabis recently highlighted a provocative possibility: classical AI systems might one day rival quantum computers in solving problems previously thought to require quantum power, such as breaking modern encryption. Drawing parallels to AlphaFold’s success in molecular modeling—an area traditionally considered out of reach without quantum simulations—Hassabis warns that AI could accelerate cryptanalysis by augmenting human researchers. This challenges the long-held assumption that only quantum computers pose an existential threat to public-key encryption like RSA. The significance lies in AI’s potential to sift through vast academic literature and optimize approaches, potentially uncovering novel attack vectors without needing quantum hardware’s raw processing. While AI currently lacks the precision to independently crack cryptographic keys, its role as a force multiplier for cryptographers turns it into a rising threat. Unlike quantum computing, which still faces substantial technical hurdles, AI systems are rapidly evolving and widely accessible, making practical cryptographic risks more immediate. The intersection of AI and quantum computing could further expedite vulnerabilities, emphasizing that encryption migration can no longer be deferred. Given these developments, the push for quantum-safe cryptography is urgent. These next-generation algorithms are designed to withstand both quantum and AI-driven attacks. With estimated timelines for industry-wide adoption stretching over a decade, immediate action is critical for governments and businesses to protect data security. The era of quantum-safe encryption is not looming—it is already beginning, underscoring a paradigm shift in cybersecurity priorities.