Anthropic’s Claude AI has uncovered weaknesses in encryption technologies designed to protect online data, highlighting AI’s growing role in cybersecurity research.
The company said the discoveries could help researchers strengthen digital security, but they do not currently threaten widely used systems.
Claude AI Identifies New Security Weaknesses
Anthropic’s Claude Mythos Preview helped researchers discover improved attacks against cryptographic methods used to secure sensitive information.
The AI system found a stronger attack against HAWK, a candidate encryption method developed for the post-quantum era. Researchers also used Claude to improve attacks against a reduced version of AES, one of the world’s most widely used encryption standards.
Anthropic emphasized that neither discovery creates an immediate risk for production systems or currently deployed encryption.
AI Takes Bigger Role in Cybersecurity Research
The findings show how artificial intelligence is moving beyond traditional software development tasks and entering advanced mathematical research.
Instead of only identifying coding errors, AI models are increasingly being tested for their ability to analyze complex systems, discover weaknesses, and assist security researchers.
Anthropic said the work demonstrates how AI could accelerate cryptography research by helping experts test stronger defenses before vulnerabilities become real-world threats.
What It Means for Digital Security
Encryption protects everything from online banking and private messages to business data and digital infrastructure. As computing technology advances, researchers continue searching for stronger methods to defend sensitive information.
The rise of AI-powered security research could reshape how companies and governments approach cybersecurity, especially as new encryption standards are developed.
What Comes Next
Anthropic’s research adds to growing interest in using AI as a tool for strengthening digital defenses. Future developments will likely focus on combining AI capabilities with human expertise to identify vulnerabilities earlier and build more resilient security systems.