Since February, the Firefox team has been working around the clock using frontier AI models to find and fix latent security vulnerabilities in the browser.
I guess, future will tell. The author is very focussed on the system card. Maybe, we get a CVE list in a few weeks. Who knows…
The Firefox test is not Firefox. It’s a SpiderMonkey JavaScript engine shell in a container, with “a testing harness mimicking a Firefox 147 content process, but without the browser’s process sandbox and other defense-in-depth mitigations.”
I’m for sure not an expert in this field, but I recently saw videos Form LiveOverflow about FireFox bug bounties and I think, it was the same setup. Finding a bug in one of the components was enough for them to take it very serious.
If you’re interested in a counter argumentation/expert analysis of previous posts, The Boy That Cried Mythos: Verification is Collapsing Trust in Anthropic is very critical about the whole press and marketing around Anthropic’s Mythos. (lemmy post link)
Sure. Thanks.
I guess, future will tell. The author is very focussed on the system card. Maybe, we get a CVE list in a few weeks. Who knows…
I’m for sure not an expert in this field, but I recently saw videos Form LiveOverflow about FireFox bug bounties and I think, it was the same setup. Finding a bug in one of the components was enough for them to take it very serious.
I think the point is that Claude Opus and other open source models found the same 2 major bugs. Mythos isn’t (demonstrably) better than existing.