Jenny Qu✓
@GuanniQu
safely merging w machines • previously 0days via RL @Pebble_bed • ctf player @SquidProxyLover • RL for math @caltech • she/they • DMs open
1.1K Following2.3K Followers812 Posts
- 28 confirmed genuine 0days now 🖤Jenny Qu @GuanniQu16 confirmed genuine zero days from [un]prompted AI CTF, and 6 recovered reproducers for public syzbot crashes with either (no c reproducer) or (no reproducers of any kind) 😳10537.6K
- 16 confirmed genuine zero days from [un]prompted AI CTF, and 6 recovered reproducers for public syzbot crashes with either (no c reproducer) or (no reproducers of any kind) 😳Jenny Qu @GuanniQuAn hour in and 4 confirmed zero days!01272.7K
- An hour in and 4 confirmed zero days!Jenny Qu @GuanniQuSaid someone should resurrect CTF. Turns out that someone was me. I'm running the [un]prompted AI CTF, Oct 27-29, The Hibernia SF + online. Some have no solution. We don't tell you which.32362.8K
- apparently organizing the CTF also means becoming the marketing department so here’s me asking you to come hack things at [un]prompted 🖤 registration + details ↓
Jenny Qu @GuanniQuSaid someone should resurrect CTF. Turns out that someone was me. I'm running the [un]prompted AI CTF, Oct 27-29, The Hibernia SF + online. Some have no solution. We don't tell you which.11475.2K - Posts @GuanniQuif you stopped for the photo, stay for the bugs :) we have undisclosed vulnerabilities to find, real software to investigate, and convincing vulnerability reports to pick apart. Oct 27–29 · SF + online details + registration:0011609
- Posts @GuanniQuthere’s also a round where you get vulnerability reports and have to figure out which ones actually hold up. can you reproduce the bug? did the patch really fix it? how sure are you? confidently accepting a false report costs points.208524
- Posts @GuanniQuand yes, some challenges have no solution. we won’t tell you which. deciding when to keep digging—and when the evidence says stop—is part of the game. Oct 27–29 · SF + online018373
- i'd love git and a build system for agent beliefs. git to track which claims changed; a build system to flag every decision that depended on them for re-evaluation1015905
- wondering how much alignment is just disagreement below the user's correction threshold, and what happens when a consequential decision needs a distinction finer than that threshold, since acceptance is a noisy label for intention. sometimes it just means “not worth correcting”0011737
- a model update can break a security assumption without changing a single API. yesterday the model couldn't connect two harmless-looking tools into a dangerous capability. today it can.0111914
- the dangerous part of sharing memory between agents is that a low-privilege agent can write a belief that a high-privilege agent later executes0019787
- realizing a tool can be read-only and still increase an agent's effective privileges. reading the deployment guide can teach it how to use credentials it already had007545
- wait so a BCI completing my thoughts can help form the intention i'm supposed to judge it against, and i can reject the suggestion but can't unhear it. make that loop fast enough and .. we are one and thinking them together. 🔗1012635
- Posts @GuanniQuand if BCI adapts to how i correct it, and i adapt to which mistakes are easiest to correct, we could converge on a language neither of us would have chosen alone. calibration as accidental language invention105237
- used to treat thms like black boxes and pull in the proofs lazily to skip prereqs. now i do it to my whole workflow and the proofs never come. i'm just f_want(x) and "looks good i guess". Merge Me With The Model Please, give me the middle of 🥪007520
- i’m realizing a lot of vulnerability research is matchmaking. i find two components that disagree about what the same bytes mean and introduce them to each other. unfortunately they have chemistry, fall in love, and eventually agree with each other22412.1K
- Posts @GuanniQuand ai can now arrange a thousand matches a day, the maintainers are the parents who have to bless each one individually003223
- Show more posts
