treadful

“Normally what would happen… is like one cop would have heard that, flip on the blue lights, race down there… and someone’s about to die,” [Flock CEO Garrett] Langley explained, describing a case in which a woman called 911 on a man, who later turned out to be holding a lighter, not a gun, as initially suspected. “Instead, they launch the drone.”

Langley claimed that “we have thousands and thousands of stories like that where communities are safer now with our drones. They’re safer with our cameras.”

That's your prime example of a community becoming safer? The cops droning a mf because he lit a cigarette and some Karen panicked? You're the CEO of the fucking company with like the worst PR and you say this shit.

Best day ever.

Unironically.

Does anyone get anything from "AI says" articles?

Did the Zippies not come?

I can't remember the last time I did. Have one in mind?

I was a dmoz editor for a bit. They created the directory that Yahoo used on their site. Kind of wish directories came back.

The metaverse predates Facebook's interest in it.

The flaws identified do not concern a cryptographic standard currently in use today, which means that modern banking and communication systems are not subject to immediate potential intrusions from A.I. Instead, Anthropic’s technology cracked a watered-down version of an algorithm for Advanced Encryption Standard, or A.E.S., a ubiquitous protocol that safeguards web traffic, wireless networks, data storage and more.

So not exactly tough-to-crack, eh?

Mythos devised the cryptographic attack on A.E.S. almost entirely autonomously, Anthropic said, but only after first refusing to contemplate the problem because it believed it was impossible to improve on existing methods of analysis. But after some coaxing, the chatbot sat with the puzzle for about a week before engineering its novel attack. Two human researchers then worked for nearly a month to verify that the method appeared correct.

A week of processing? Wish I knew more about these systems. I wouldn't expect there's any way a single prompt-to-response took a week. Maybe it's some complex "agentic" setup that broke the problem out into parts, then worked on each part, validate it, then moved on. But that's pure speculation on my part.

You type like an AI user.

the agent was operating on instructions that reduced cyber refusals to successfully evaluate the model

I can try and rephrase: It's shocking to me how many people support invasions of privacy for the purposes of detecting and prosecuting ideas and thoughts.