One slide in that day’s presentation showed that in April alone, Mythos had uncovered 90 “critical” bugs and 141 “important” ones in SharePoint, Microsoft’s widely used collaboration software. In the first half of May it found even more.

If an organization the size of Microsoft is struggling to fix 231 bugs in a month, that's an indictment of Microsoft, not an example of some impossible task. Tech companies have been laying off people left and right and this is a threat to their core business. Most of these shouldn't be more than adding some input checking.

Microsoft: "Why is our AI so inferior? We trained it on the entire private code base for Windows and Teams!"

But.. but... I thought closed source was more secure!

Aren't microsoft products half vibe coded at this point anyway? Look at all these fires I put out, says fireman with trunk full of flares and gasoline.

This whole software bug spiel is to convince everyone that it is unsafe to code without using an LLM. Effectively forcing companies to use LLMs.

That's why you have AI to fix the bugs as soon as you find them. Unless those LLM are just not as good as advertised. Then it's a strange message to put out there.

Fuck Anthropic. They buy used books and put them through a shredder to train Claude.

Microsoft being slow to fix bugs isn't news.

this type of thing really shows llm's to me. its great at creating this busy work but not so much with fixing it. so it can find them and we know microsoft is all in on ai coding, but it does not seem to help fixing them quickly. they are spending so much money to muck things up more and make more expense.

AI slop kettle calls AI slop pot black.

While these models can find more bugs, it seems that these bugs do not convert to exploits to a higher rate than before

The Open Source epidemic, people.

The Epsteins and their Corps seem to be organizing/manipulating the information sphere to create fear/hype around their top models. It seem to be centrally coordinated and different US tech Corps/players enters and leave. We keep getting either exaggerated/fake or 'woopsie' security narratives thrown at us.

The normal purpose is to create fear/anger before a false-flag operation, so we'll likely see many claims about AI-hacking and a single great hit (the flag) against the Merican people, that will point towards one of the nations that the US/Epsteins have manufactured consent to attack - likely Iran. Old playbook stuff with a new topic..

Wild that so many people are still stuck believing the information from the Epstein class' owned information sphere.

..on the other hand, I wouldn't want to BE the Epstein class when ~250M Mericans finally discover they got mind-raped for their entire lives..

LLMs probably created a lot of these bugs in the first place.

We're talking about Microsoft here so I think it's probably found three bugs, two of which were already reported.

Can anybody help me understand the threat of these advanced models once the public and adversaries have access to them? I understand they can find software bugs and vulnerabilities and I get that US tech companies are using it to make their software more secure, but all of these companies have the benefit of their source code being available as an input to the model. Everybody else only has compiled binaries. Do the bad guys use the models to decompile or reverse engineer the code first?

Sounds like Microsoft needs to hire some people...

In the end, LLM’s will find and fix those same bugs autonomously, people will do the final sign off and software will be simultaneously more secure and less secure as software continues to be tested worldwide wide