Getting access to Mythos high on priority but other AI models also being utilised: IT Secy
Getting access to Mythos and other advanced frontier models is high on the government's priority, and while talks and processes for the same are on, alternate AI models are also being leveraged for now to track and fix vulnerabilities, IT Secretary S Krishnan said on Monday.
Mythos - an advanced AI model developed by Anthropic - has been able to "lay bare" vulnerabilities that existed in the original code of widely-used software programmes but were often left unresolved because they were considered low priority, too costly to fix, or had simply gone undetected, he noted.
"What this gives is an opportunity now to actually identify many of these vulnerabilities and correct them. Clearly, getting access to Mythos and similar advanced frontier models is very high on the priority of the government, and this is something that we have taken up with our counterparts in the US and with the respective companies," Krishnan said.
There is a process involved, Krishnan said, adding, "We are negotiating that process; we are going through that right now."
Even as the negotiations continue, the government has created a "sandbox" environment using alternative AI models with an estimated 60-70 per cent of the capabilities of the most advanced systems. The objective is to test code, identify vulnerabilities and develop secure workflows before access to more powerful models becomes available.
"You know what the stand of the US currently is...that there are export controls on products of this nature. What we are instead doing to prepare ourselves is that CERT-In themselves has a sort of a sandbox or a 'war room' where they have other models which they are using...which they believe at about 60 to 70 per cent of the capability of what Mythos can do," he said, adding that various elements of code are being probed, vulnerabilities are being identified and fixed.
Mythos - an advanced AI model developed by Anthropic - has been able to "lay bare" vulnerabilities that existed in the original code of widely-used software programmes but were often left unresolved because they were considered low priority, too costly to fix, or had simply gone undetected, he noted.
"What this gives is an opportunity now to actually identify many of these vulnerabilities and correct them. Clearly, getting access to Mythos and similar advanced frontier models is very high on the priority of the government, and this is something that we have taken up with our counterparts in the US and with the respective companies," Krishnan said.
There is a process involved, Krishnan said, adding, "We are negotiating that process; we are going through that right now."
Even as the negotiations continue, the government has created a "sandbox" environment using alternative AI models with an estimated 60-70 per cent of the capabilities of the most advanced systems. The objective is to test code, identify vulnerabilities and develop secure workflows before access to more powerful models becomes available.
"You know what the stand of the US currently is...that there are export controls on products of this nature. What we are instead doing to prepare ourselves is that CERT-In themselves has a sort of a sandbox or a 'war room' where they have other models which they are using...which they believe at about 60 to 70 per cent of the capability of what Mythos can do," he said, adding that various elements of code are being probed, vulnerabilities are being identified and fixed.
Next Story