The cURL creator cares more about his code than most companies do, I would conjecture; while this is a good reality check, I expect that it will be far more useful in the programming sweatshops responsible for the Domino's website et al.

It's still the wrong tool for the job. The sheer number of things generative AI is used for is embarrassing.

But after the system turned up just a single vulnerability, he concluded the hype around Mythos was “primarily marketing” rather than a major AI security breakthrough.

Now there's a surprise.

If it's all hype, then how come Mythos found 150 bugs in FireFox?

Personally I am responsible for some corporate software that Mythos found bugs in. Real ones. We're fixing them literally RN. Not caught by formerly industry-standard static and dynamic analysis tools.

Good for cUrl if it had few flaws. Be aware you're like a person who was already masking and distancing who didn't get COVID. Your good, prophylactic work does not mean others are not severely impacted or that it's "just a flu".

I said it at the time and I'll say it again: the only innovation that Anthropic developed for Mythos is a system card that performs regulatory capture faster and more completely than any software documentation ever written. I'm not impressed by Mythos, but I am terrified of Anthropic's technical writing team.