OpenAI and Anthropic both claim AI cracked a code-security puzzle

OpenAI says its AI models disproved a foundational idea in cryptography, the science of secret codes. The company made the claim on Monday, according to The Information. The outlet reports that Anthropic, another AI company, made its own cryptography breakthrough claim the same day.

What was claimed

Cryptography is the science behind keeping information private and secure online. A foundational idea in the field is one that other work is built on. OpenAI says its models showed one such idea to be wrong.

OpenAI also announced two other breakthroughs, one in physics and one in mathematics. Earlier this month, it released more than 700 math papers produced with its AI models.

What is not known yet

• Which cryptography idea OpenAI says it disproved.
• What Anthropic actually claimed, and how it compares.
• Whether outside experts have checked either claim.
• Whether the findings affect any security tools people use today.

These are company claims. They have not been confirmed independently, at least in the material available. Claims like this usually need review by outside experts before anyone treats them as settled.

For now, nothing here tells everyday users or business owners to change how they protect their data. The story is worth watching because it hints that AI could one day help build and test online security.