India Tests Open Source AI Models to Measure Cyber ​​Bugs as Threats Escalate | Today’s news

“What we do to prepare for AI risks is our own Cert-In sandbox for such vulnerabilities, in which the body tests cyber vulnerabilities on popular platforms. These open source models and others can produce about 60% of the total performance that (Anthropic) Mythos is capable of, and we use them to identify different systems that are patched. Mythos has become widely available, we should already have a more secure enterprise environment,” he said. S. Krishnan, secretary of Meity, at a press conference on Monday.