> Overall, the threat actor was able to use AI to perform 80-90% of the campaign, with human intervention required only sporadically (perhaps 4-6 critical decision points per hacking campaign). The sheer amount of work performed by the AI would have taken vast amounts of time for a human team. At the peak of its attack, the AI made thousands of requests, often multiple per second—an attack speed that would have been, for human hackers, simply impossible to match.

Weird flex but OK

yeah but the link states "The 20 million user conversations were randomly sampled from Dec. 2022 to Nov. 2024" so this makes no sense. 2024 was much longer than 30 days ago

Because the court ordered them to retain the records longer than they normally would.

Most of Apple's business model is built on the back of this resource bloat

I use Claude Code, haven't used Codex yet (should I?) - but in Claude code you can spin up sub-agents to handle these big refactors, with the master context window just keeping track of the overall progress, bugs, etc and providing instructions to the subagents to do the rote work.

IMO yes. It is less polished but IMO the model is way better. I moved over from claude completely and cancelled my max subscription. Less polished, slower but the results are better and you have to do less steering

Not really. Apple is a very strong competitor here.

The inch thing is because it's a screen: all screens are measured in inches pretty much everywhere, even in the EU

And I hate that. Wish screen sizes were measured in centimeters.

Yep, inches won for screen sizes.

Aaaand for various fittings for plumbers.

yeah I found a setting in Brave to block them the other day and life is much better since

"We are now confident we know how to build AGI as we have traditionally understood it." - Sam Altman, Jan 2025

'as we have traditionally understood it' is doing a lot of heavy lifting there

https://blog.samaltman.com/reflections#:~:text=We%20believe%...

https://updog.ai/status/openai issue history looks terrible. Wonder how you ping openAI for this; with a completion attempt on a particular model?

It is based on the impact on Datadog's customers, not on synthetic queries / pings

I'm also inclined to overthink this stuff, but please try not to worry so much. You were outdoors, wearing a mask, for one day. This is very different from occupational exposure with no protection over decades.

There are a thousand things that can increase your risk of cancer by 0.001%. Heart disease kills >20X more people than asbestos related illnesses in the US. Focus on being healthy and happy.