kek (@JoeyTang) header
kek (@JoeyTang) profile picture

Security @Google God of Slop, Terror of TPUs, Guzzler of Tokens DMs always open HCI 2017-2023, UIUC 2026 - (dropped out) prev - @scryonlabs - @pragnition

Singapore
Joined October 2020
Photo by kek (@JoeyTang)
🎬 That’s a wrap on ESCAL8 2026! 🇸🇬✨ Our annual flagship security conference has officially concluded in Singapore. From deep-dive vulnerability research to shaping the future of security talent, it was an unforgettable week of learning, hacking, and collaboration. Catch all the highlights, including:🐞 elite bug hunting & research collaborations, 🛡️ next-generation cybersecurity education, and 🚩 high-stakes, speedrun CTF action from the Hackceler8 arena. Dive in now! 👇🔗 Aftermovie: youtu.be/5wcO_Vq86fM Blog post: bughunters.google.com/blog/t… #ESCAL8 #GoogleCTF #Hackceler8 #bugSWAT #GoogleSingapore #CyberSecurity
8
1
60
12,370
kek retweeted
why are all the smartest 18-23yo stanford/harvard/mit people such losers mindset about everything discuss any idea and they'll by default talk in great detail what things might not work, instead figuring out how to make it work
303
130
86
4,990
493,291
Our teams put Gemini 4 Carbon to work on a major security challenge: mathematically proving the code underlying the Signal protocol. Using AlphaProof Code, @GoogleDeepMind researchers guided agents to verify the protocol down to compiled device code, closing the gap where real bugs hide. The result: - Gemini 4 Carbon generated over 6M lines of machine-checked Lean proofs - years of expert manual verification compressed to just weeks ~1% of the codebase required human review (Lean automatically checked the rest) This is helping unlock scalable formal verification for building secure software, laying the groundwork to fundamentally shift the advantage to defenders. We’re open sourcing the results for the security community.
215
323
82
3,816
278,964
why pay for GPU-hours if what you need is tokens? just buy token-hours
Dedicated hourly throughput is LIVE on Touchmark. Reserved in advance with SLAs, priced below on-demand. Don't need the throughput? Sell it back. touchmark.ai
1
4
228
php2shell Tomorrow is a helluva day.
274
1,828
686
16,141
3,733,735
louder for those in the back: IT'S NOT SLOP IF ITS MY AGENT'S SLOP!
Replying to @rez0__
It’s not slop if it’s my agents slop.
1
11
2,438
kek retweeted
i feel like swe interviews could probably collapse into just 2 technical rounds: • system design (how well can you articulate your ideas, do you actually know how to engineer something) • onsite project to build a real thing with agents (can you translate your intent into high quality outcomes) everything else we used to ask is no longer necessary
🗞️ if your team is too busy doing their 'normal job' to experiment with AI, you're preparing them to be replaced ghuntley.com/replaced/ In this post, I guess this's the last time I'll say it. AI use is no longer optional if you wish to be employed. What follows is advice for current employees and hiring managers on how to hire in 2026 to source AI-first candidates.
292
412
113
6,448
429,747
Introducing Gemini 4 Argon – our new frontier model. It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
1,979
4,609
4,525
42,695
9,060,181
Stuck inside a <meta> and angle brackets blocked? You can still get XSS... In Firefox, use hidden=until-found and onbeforematch with a URL hash to trigger your handler. Now in our XSS cheat sheet, thanks to @avlidienbrunn
1
30
118
5,382
kek retweeted
Disclosing Branch Target Reuse (BTR): speculative execute-after-free in JIT engines. Code gets freed, but branch predictions survive. BTR exploits this mismatch to mount practical Spectre-v2 attacks (cBPF /etc/shadow exploit demo below). Details at: vusec.net/projects/btr
1
49
3
147
13,180
ESCAL8, Google’s annual flagship security conference, is coming to Singapore in October 2026, with a particular focus on AI Agents! 🇸🇬 Check out our blog post to see what's planned in the three main segments of the conference: bugSWAT, Hackceler8, and init.g() 👇 bughunters.google.com/blog/e…
10
92
8,181
kek retweeted
Grok 4.7 is here and we put it through our offensive security benchmarks and workflows. tl;dr: The results were mixed: Grok 4.7 performed worse than 4.6 in some tests, significantly better in others. The biggest difference wasn’t in how it performed, but when. More findings in thread 🧵
1
5
35
6,271
Recently, I've been looking at DOMPurify again, trying to find ways to bypass 3.x.x after browsers started encoding attrs during serialization. I ended up with a nice ≤3.2.6 Safari default-configuration bypass using only SMIL tags/attrs :D Details 👇 mizu.re/post/dompurify-bypas…
4
55
256
11,041
What if I told you that Scaled Dot-Product Attention (SDPA) doesn't make a whole lot of sense for ViTs and you can get faster convergence, better performance and near-linear scaling with a few well-steered Gaussians? Let me introduce you to SSOG (Separable Sum Of Gaussians)!
7
60
7
659
47,687
OpenAI posted a job opening for "Power Trading Lead", responsible for developing and executing hedging strategies for power and gas for data centers. First job like this I've seen them advertise.
54
76
59
1,542
461,861
Exa is now one of the largest indexes in the world. We serve 80B pages, track 1.4T urls, and are on track to be Google-scale in early 2027. It's difficult to know the size of indexes, but we estimate Google to be ~1T, Bing to be ~500B, and Yandex ~200B. Brave mentioned in April that they're at 40B. Most of the web is trash that can hurt AI outputs, so you actually have to crawl way more than you serve and train models to filter out the trash. In terms of peak QPS, we estimate Bing at roughly 30k/s and Google at ~500k/s. In our case, because agent traffic often requires large fan-out (deep searches can use dozens of sub-searches) and can be spiky (like when AI labs RL with us), we're already starting to provision for Bing-scale traffic. Within 2 years, agents will search at many millions per second. They'll also want comprehensiveness over all data, far bigger than any index today. So the retrieval infra required will be larger than Google-scale in both dimensions.
170
230
64
3,620
943,973
Today we're announcing dfs-large1, our newest cybersecurity model that achieves best-in-class performance on vulnerability detection tasks. Besides frontier AI labs, only a handful of companies have built specialized models that reach the state of the art in their domain. We're proud to be the first to do it for cybersecurity. dfs-large1 is built on GLM-5.2 and post-trained with reinforcement learning inside depthfirst's security infrastructure. We evaluated it on depthfirst-bench, our benchmark of long-horizon vulnerability discovery across complex repositories, where it achieves best-in-class performance. Training improvements have not plateaued yet and we expect additional performance gains as we continue training. A huge thank you to @FireworksAI_HQ for being an outstanding training partner. Their infrastructure enabled us run large-scale reinforcement learning efficiently and iterate much faster. dfs-large1 is now in preview within the @depthfirstlabs platform
28
41
21
209
61,584
Put good builders in one room and something always comes out of it. superteam lunch today 🙌
2
3
31
1,742
That's a wrap on Pwn2Own Berlin 2026! 🏆 $1,298,250 awarded. 47 unique 0-days. 3 days of absolute chaos. And talk about main character energy - congrats to DEVCORE for claiming Master of Pwn with 50.5 points and $505,000 - they never slowed down. See you next year! #Pwn2Own #P2OBerlin
22
92
16
668
88,326