Google DeepMind says AI agents found two flaws in 1,024 bit RSA
The company says the two flaws are not in RSA systems with 2,048 bits.
Google DeepMind says that AI agents, with humans in control, found two flaws in RSA encryption with 1,024 bits. The company says this shows that AI can make tests of encryption faster. It says the two flaws are not in systems with 2,048 bits. The source gives no test results and no data on the two flaws.
What is new
Humans guided the AI agents as they tested RSA systems. The agents found two flaws. The company calls the work an example of faster tests of cryptographic systems.
The proof
The source gives no test results, no method and no data on the two flaws. One number limits the result: 2,048. The company says systems of that length are not in the group.
What is not known
The source does not say how much faster the tests were. It does not say how much work the humans did. The company also says that cryptographic standards must have more tests as AI advances mathematics and computing.
Insight: An AI company now tests the locks of the web
Google DeepMind used AI agents to test encryption, with humans in control. An engineer at Google said before that a human examines each code change that AI writes. This reads as a race to set how AI is tested. the company that makes the tools also gives the results. Leaders must find out who checks the checks.
Sources
Posted
