If true, this says less about Mythos, and more about how the security posture of those systems is so awful that being airgapped is really the only thing saving them.
ACSC found in their testing had fundamental security controls like network segmentation work well to slow down AI driven adversaries.
Can attest to this. The door going into the secure area is the most secure part of a SCIF and always has been. Some companies do the LAN inside the SCIF a bit better than others, but generally its not significantly any different from what is otherwise corpnet on the outside, and tends to be weaker because there (not so rare to find) isn't a corpnet on the inside.
As the article mentions, no new attack vectors were found, but just the scale of it is the concerning part.
I am highly skeptical of this. Show me a properly [redacted] and approved video of the bots in action at the NSA after it is cleared for public display, hosted on nsa.gov or it didn't happen.
If true, this says less about Mythos, and more about how the security posture of those systems is so awful that being airgapped is really the only thing saving them.
ACSC found in their testing had fundamental security controls like network segmentation work well to slow down AI driven adversaries.
https://www.cyber.gov.au/about-us/view-all-content/news/fron...
Can attest to this. The door going into the secure area is the most secure part of a SCIF and always has been. Some companies do the LAN inside the SCIF a bit better than others, but generally its not significantly any different from what is otherwise corpnet on the outside, and tends to be weaker because there (not so rare to find) isn't a corpnet on the inside.
As the article mentions, no new attack vectors were found, but just the scale of it is the concerning part.
Fairly certain this is over a week old, not been verified since, and has been challenged by multiple people.
I am highly skeptical of this. Show me a properly [redacted] and approved video of the bots in action at the NSA after it is cleared for public display, hosted on nsa.gov or it didn't happen.
I can find absolutely nothing in mainstream press to back up this claim. Does anyone have a reputable source?
didn't they have weeks of advance use to patch? Goddamn
Did this current administration negotiate a pre-IPO allocation stake with Anthropic for that scare story?