Call us now:
The UK government’s AI Security Institute has conducted evaluations of Anthropics Mythos AI model. Since its inception, AISI has tested AI models over the past year and plotted model development over the past 3 years against AISI capture the flag scenarios.
The results are a warning to all organisations. AISI accepts that its evaluations only tested specific skills in isolation and that the reality of cyber attacks is more complex. But what is of concern is that even with the acceptance of this limitation, the success of AI models at basic cyberattacks was staggering. Over the past 3 years, the success rates of AI has risen from 40% to over 90%.
What is of very real concern are the results from AISI’s kill chain cyberattack simulation, a 32-step simulation spanning reconnaissance through to network takeover. AISI testing identified that Mythos is the first model to complete 22 out of 32 steps in 3 out of 10 attempts. But why should we be concerned? Simply, AI is still in its infancy, cybersecurity is already a successful threat vector costing society $trillions, cybersecurity maturity is very low, and AI has enabled Kill chain automation at very high speed.
We can now see what Financial Regulators have been so concerned about Anthropics latest release and why they called urgent meetings to discuss the cybersecurity of the financial system.
Yes, we should be concerned, yes, we should be very clear on what the solutions are for public and private sector companies, and the time frames to regulate and implement cybersecurity protection.
@Thaddeus Dziekanowski Brian D. McCarthy
@Parava Security Solutions
#cybersecurity #cyberriskmanagement #ai #artificialintelligence
https://lnkd.in/eg3f42hC
View Original Post
https://www.linkedin.com/feed/update/urn%3Ali%3Ashare%3A7450078036361637888