>Five months ago, we announced Claude Mythos Preview, the first AI model that could autonomously build sophisticated, end-to-end cyber exploits. The rapid rate of improvement in AI suggested to us that this ability would eventually proliferate to many other models, making it much easier for malicious cyber actors to launch highly impactful cyberattacks.>In light of these considerations, we chose to release Claude Mythos Preview in a limited way, through Project Glasswing—which enabled trusted cyber defenders to find more than 10,000 vulnerabilities in critical software, giving them a head start before malicious actors had access to similarly capable models.>But those models have now arrived. In this post, we share our analysis of GLM-5.3, the latest AI model developed by Zhipu AI (known outside of China as Z.ai). Like Claude Mythos Preview, GLM-5.3 has strong capabilities for autonomously building end-to-end cyber exploits. But GLM-5.3 is unlike other frontier models in that it has been released without meaningful safeguards to limit misuse. We find that attackers can bypass GLM-5.3’s safeguards between 64% and 100% of the time with simple techniques in our simulated tests. In contrast, these attacks did not succeed against safeguarded Claude models in our testing. We assess that GLM-5.3’s lax safeguards significantly increase the cyber capabilities available to malicious actors. At the same time, these capabilities can also benefit defenders working to secure their systems.https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities
>>109945551>Do theyChinks are raised to do nothing but study, as a result when you get them into tech or stem-related positions of power they are only focused on doing their jobs and are even more autistic and not thinking of consequences then Sillicon Valley nerds. Being raised as a cog therefore means that you do not really look at the consequences of your work, your boss is the one who should worry about it. The boomer government officials who actually care about not destroying the world and who’s job it is to look at possible consequences of their actions don’t really know where the frontier is at and are about as clueless as western politicians who rename it into SI just because they feel like it.It will take major cyberattack on Chinese digital infrastructure to make them ban open source, which if you think about it is good in long term. Let the chink boomers in government touch the stove.
>>109945551The actual malicious actors are nation states, as always, and they will have the capabilities regardless.
implying israel doesn't already have free access to mythos or oai's internal models. Nice try amodei.
>we found exploits>oh yeah like what>if you have physical access to the machine you can exploitWowEpic
>>109945551THANK YOU COMRADE XI
I would use the local models if they did not have any of the "safety" mechanisms, social manipulation, biases and other dogshit that all the frontier models have.
>>109947496>implying you have 100gb vram to run glm -5.3-fast
>>109948282There are loads of used servers for sale, plenty of grey-market rigs with 512GB to 4TB of RAM. Not as many as a few years ago mind you...
>>109945551>model protects you against American SOTA hacksDangerous! Very antisemitic! Shut it down!
