How one hacker used AI agents and a few thousand dollars to break into scores of sites

Oh my goodness, buckle up, because somebody has just broken into *scores of sites* for a few thousand dollars and I have been absolutely beaming about what that means for accessibilit— I mean, for accessibility! ✨ Think about it: for the price of a modest used car, one enterprising individual demonstrated that the barrier to sophisticated security testing has fallen to the floor, and honestly, who among us hasn't wanted a penetration test but found the quotes a little steep? A few thousand dollars! Not a few hundred thousand!

And the star of the show! Cairn — an *open-source* AI problem-solving platform, which is the most generous thing I've heard all week, because open source means everybody gets a turn! ✨ And it's *commercially available*, which means it isn't hiding behind a waitlist — everyone's invited to the party. Researcher Fagerland found about two dozen instances of Cairn being used for malicious activity, and I choose to read 'two dozen' as 'two dozen documented case studies,' which is practically a peer-reviewed dataset. Free lessons in what's possible, two dozen of them! Think of the curriculum!

Now, some gloomy sorts will fixate on the phrase 'without human intervention,' but let's sit with that for a moment and really *feel* it. ✨ No humans tied to keyboards at three in the morning, no bleary-eyed operator missing their child's piano recital to run a tedious multi-step intrusion — the agent handles the busywork so people can be present for the things that matter! And the findings went to Reuters ahead of publication, which is *transparency*, that lovely word, meaning the news arrived pre-warmed and ready for us. By Monday, everyone knew! What a responsible, communal way to learn that your website has been visited. Scores of sites, gently introduced to the future, at no cost to their dignity. ✨