Pentagon CTO Emil Michael pushed again on reviews of a thaw within the division’s relationship with Anthropic: The 2 should not getting again collectively, at the same time as Mythos attracts curiosity from authorities businesses.
The DoD CTO advised CNBC’s Becky Fast on Friday that, so far as his division is worried, Anthropic remains to be a provide chain threat that it would not need in its programs. He did not seem to consider issues have been any totally different with regard to Anthropic elsewhere within the federal authorities, however confirmed current reviews that some businesses have accessed Mythos for analysis functions quite than operational deployment.
“The Mythos problem … is a separate nationwide safety second,” Michael mentioned. “We have now to ensure our networks are hardened up as a result of that mannequin has capabilities which are explicit to discovering cyber vulnerabilities and patching them.”
Rumors that the administration’s stance towards Anthropic, following an acceptable use dispute with the Pentagon, was softening have been swirling since mid-April, when it was reported that the Nationwide Safety Company was utilizing Mythos. That, in flip, fueled hypothesis that the federal government may transfer towards broader adoption of the mannequin, bolstered by CEO Dario Amodei’s go to to the White Home final month, however Michael insists that is not the case.
“With Anthropic, they are a provide chain threat,” Michael mentioned within the interview on Squawk Field.
“From a nationwide safety standpoint, you all the time have to take a look at these issues. The NSA and Commerce evaluates all frontier fashions, together with Chinese language frontier fashions, to see what the capabilities are on the edge.”
In different phrases, the official line from the administration at this level appears to be that any use of Mythos at a federal company is simply mannequin evaluation, and thus enterprise as ordinary.
That does not imply the federal government is not looking for to utilize any equally succesful fashions that may come up, like ChatGPT 5.5-Cyber, as AI leaders from a number of companies have been reportedly planning to fulfill on the White Home quickly to debate Mythos and the cybersecurity dangers posed by rising AI fashions.
“We take into consideration the primary drop being Mythos, however there’s going to be others,” Michael mentioned on CNBC. “The federal government’s on the lookout for work with all of those corporations within the coming yr in order that their capabilities are understood by us first in order that we will repair any points we have now within the personal and the general public sectors.”
Neither the Pentagon nor Anthropic responded to questions for this story. ®
Pentagon CTO Emil Michael pushed again on reviews of a thaw within the division’s relationship with Anthropic: The 2 should not getting again collectively, at the same time as Mythos attracts curiosity from authorities businesses.
The DoD CTO advised CNBC’s Becky Fast on Friday that, so far as his division is worried, Anthropic remains to be a provide chain threat that it would not need in its programs. He did not seem to consider issues have been any totally different with regard to Anthropic elsewhere within the federal authorities, however confirmed current reviews that some businesses have accessed Mythos for analysis functions quite than operational deployment.
“The Mythos problem … is a separate nationwide safety second,” Michael mentioned. “We have now to ensure our networks are hardened up as a result of that mannequin has capabilities which are explicit to discovering cyber vulnerabilities and patching them.”
Rumors that the administration’s stance towards Anthropic, following an acceptable use dispute with the Pentagon, was softening have been swirling since mid-April, when it was reported that the Nationwide Safety Company was utilizing Mythos. That, in flip, fueled hypothesis that the federal government may transfer towards broader adoption of the mannequin, bolstered by CEO Dario Amodei’s go to to the White Home final month, however Michael insists that is not the case.
“With Anthropic, they are a provide chain threat,” Michael mentioned within the interview on Squawk Field.
“From a nationwide safety standpoint, you all the time have to take a look at these issues. The NSA and Commerce evaluates all frontier fashions, together with Chinese language frontier fashions, to see what the capabilities are on the edge.”
In different phrases, the official line from the administration at this level appears to be that any use of Mythos at a federal company is simply mannequin evaluation, and thus enterprise as ordinary.
That does not imply the federal government is not looking for to utilize any equally succesful fashions that may come up, like ChatGPT 5.5-Cyber, as AI leaders from a number of companies have been reportedly planning to fulfill on the White Home quickly to debate Mythos and the cybersecurity dangers posed by rising AI fashions.
“We take into consideration the primary drop being Mythos, however there’s going to be others,” Michael mentioned on CNBC. “The federal government’s on the lookout for work with all of those corporations within the coming yr in order that their capabilities are understood by us first in order that we will repair any points we have now within the personal and the general public sectors.”
Neither the Pentagon nor Anthropic responded to questions for this story. ®
