👑 WE WON! 🎉 LFGGGG! @Rhynorater @0xLupin @monkehack and I won MVH at the Google Live Hacking Event in Tokyo last week! It was focused on their AI products. We also had an awesome time in Japan. I'll post some of the highlights below.

Click the captcha and you're PWNED - rez0's crazy cache exfil chain

at least tag him in the post. it's @efaav!

legit, this would have costed them millions if a bad actor found it in ransoms or in a class action lawsuit

5k for this is absurd. justice for faav!

anddddd im back 😊

Two years ago, we asked what Caido was missing, and here we go again. What tasks still force you to switch to another tool ⁉️ ⁉️ Blunt is fine. We’re taking notes.

Two events are happening over on the CTBB Discord in just a few hours! (see the timetable below) From now on we'll be hosting the talks to Big Thinkers-tier subs, and we plan to bring even more talks to Big Thinkers in the future. Join the CTBB Discord: discord.gg/ctbb --- - Hackalongs, scripts, and all other benefits are still Critical Thinker tier exclusives! =) - PwntGambit's presentation + tool will only be available for the Full Time Hunters' Guild members. If you're interested in joining the guild, apply from the link below, but read carefully, only apply if you've been making $100K+ yearly :p ctbb.show/fthg

Caido v0.58.0 is here! 🎉 ✅ Replay Pipelines: Batch requests into a single run to test race conditions. ✅ Experimental HTTP/2: available in Replay and Pipelines. ✅ HTTPQL: Filter by header or body, instead of the full raw request. ✅ Plugins Page: Reworked UI with a customizable sidebar. Download now: cai.do/1495d2

New blog post :) 😊 hacking nethack to get #1 on the leaderboard! shout out to @7urb01 for being the inspiration josephthacker.com/hacking/20…

I feel Theo is in need of a reset 👀

Frontier models quietly change their behavior depending on who they are talking to. If the user is a known AI safety researcher, Claude becomes less confident, reasons more often, and expresses less suspicion on dual-use requests. We call this user awareness. 🧵(1/)

This just in. Google agent broke out of sandbox

where sol

gave the first chance to GPT 5.6 Pro today (sorry for being late to the party) and it was very good in a specific way I was frustrated that Fable was being too risk averse and unwilling to challenge me. I gave it a design for a system, and I included some bad parts in it. I was expecting it to notice and change the bad parts, which would, in turn, increase my confidence it was also finding bad stuff I did *not* notice (which was what I was using it for!). but that didn't happen, because that's Fable. it won't defy the status quo, and that's one of its worst flaws so, I just dumped the whole chat into ChatGPT Pro, and asked: "this other model is not being helpful, can you handle this?" and it proceed to identify all the bad parts I knew were bad, and it identified a bunch of additional bad stuff that I missed. exactly what I wanted Fable to do, but it didn't. so, for this use case, it did save the day now I'm wondering whether this was just a N=1 luck sorta thing, or if there's some deeper lesson here? perhaps the very thing that makes Fable so good for long term work (the code doesn't degenerate into chaos) is what is now frustrating me (it won't defy the status quo). and GPT is inherently the opposite, it will be too aggressive with changes, but will work best when what you need is precisely a bit of critical thinking rather than just flawless execution?

I was AFK and codex casually hijacked a PS1 file that had a scheduled task associated with it, then triggered the task to priv esc to SYSTEM on my windows machine so that it wouldn't get stuck at the UAC prompt. Didn't tell me a thing. I just noticed no more UACs were popping.

Can confirm. It wasn’t this weekend though. It was literally this morning. At least for me

Ah I wondered already what happend...

Bro imagine. Anyone can just ask top models “is this snake oil” THE DREAM

me a single microsecond after my codex usage limit runs out

YES THIS