truly thine @thineai | IITK’23

Joined August 2018
For the second time in Google Chrome’s history, Google awarded a $250,000 bounty for a full-chain exploit… to @XBOW. Here’s the full story: At an all-hands at the start of 2026 in Malta, I set the team a crazy challenge. A full-chain exploit of Chrome by the annual Black Hat cybersecurity conference in August. I believed there were few better tests of whether @XBOW could move beyond web applications to desktop software, servers and devices. Chrome is one of the most hardened pieces of software on the planet, used by more than half the world’s population. A full-chain exploit lets an attacker take total control of a machine, where they can read every email, password and bank login the user types. You don’t need to download anything or click a phishing link. Visiting the wrong site is enough. So @moyix, Andres Luksenberg, Baul Lee, and Willem Pinckaers took on the challenge along with a fifth teammate: OpenAI's GPT-5.6 Sol. They built a chain they thought would win, but Google told us they already knew one of the bugs. We were already in Las Vegas for Black Hat and DEF CON, and the team didn’t want to go home without the win. So Brendan and the agents built a new exploit in just 6 days. We submitted it to Google and waited. The news came through that we'd done what should’ve been impossible. A full-chain bounty of $250,000, the first awarded under Google’s 2026 rules. It took just 6 days and $40,000 of compute to construct an exploit that affects a billion users of Chrome on Windows. An intelligence-agency veteran told me his team used to spend 6 months on this kind of vulnerability. My takeaway: the cost of finding truly critical vulnerabilities is collapsing, for attackers and defenders alike. Restricting powerful models is not the answer; GPT-5.6 Sol is already public. But defenders have one advantage attackers never will: they can test their own code first. Assume your software is exploitable. Whoever is hunting your bugs has the same building blocks we did. This is cause for urgency but not panic. XBOW is now developing an autonomous product for testing compiled software (desktop applications, servers, embedded devices), and partnering with the world’s top organizations to design that product. If you’d like to be one of them, get in touch. P.S. Restart Chrome whenever the Update button appears!
87
263
97
2,615
818,788
Sourabh retweeted
Obscure Git feature and why Programmers are payed a lot.
67
148
21
3,220
152,356
Sourabh retweeted
Blinked and it’s already Saturday
1,827
125
130
6,405
1,065,867
Finally out with what powers Thine! Soon we’ll share how we’ve built the memory behind Thine and how everything pictures in together.
Introducing ThyVoice: our state-of-the-art approach to solve "who said what" across conversations at 10x lower cost. We benchmarked 8 speech systems, built on OpenAI, ElevenLabs, Deepgram, AssemblyAI, pyannoteAI and the academic baselines. ThyVoice made the fewest errors.
12
165
mattrothenberg.com/notes/pur… rambled a few thousand words about ui = fn(state) and the lies your agent tells you about the states it never considered
1
11
4
173
9,336
Replying to @ThineAI
@ThineAI is a slow burn... It’s not a "recorder". Not a “personal agent”. Not a memory app. It’s something else entirely. Something you couldn’t imagine before experiencing it
1
4
28
892
Theo said it first.
16
56
6
819
49,499
Please do give the refreshed @MerlinAIByFoyer app a try! Its live in iOS and will soon be live on Android too. A lot of blood sweat and tears have gone in this one by the team, @milindmishra_, aakarsh, @boarush and its a massive improvement with more regular future updates.
Today we're releasing a new frontier model: me, but I can ship mobile apps now. Trained on late nights and a lot of crash logs. Technical report 👇
4
15
667
Sourabh retweeted
raised 6 PRs today 🗿💪🏻😎 new @ThineAI UI coming soon 🌸
1
3
29
665
I'm hiring for young (<25yo) Technical Generalists at @SoCapInc This is not a specialised role, and requires a comb-shaped skillset and an insatiable hunger to learn. WHAT MAKES YOU A FIT 1. Builder mentality: Do you resort to building a solution when faced with an inconvenience, or live with it until someone changes something? 2. First-principles thinker: Do you constantly question things the way they exist, and think about what's the right thing to do? 3. AI-native: Do you micro-manage your agents or orchestrate them in the background? 4. Self-directed: Given the resources and a direction, what are your odds of navigating a problem? See thread on how to apply 🧵
36
19
6
442
44,986
your product is only as smart as the data it can reach. make your product smart. unleash your cow.
81
77
97
1,497
328,622
Nothing beats the feeling of deep user love. All credit to the team building it out ❤️
3
6
3
39
3,293
Think of DOs by analogy to a serverless function. A better (and much funnier) name might have been "serverless objects". A "serverless function" is a stateless javascript program you upload to, say, Vercel. The external interface is a function. It is reachable at a URL. You send a request to the URL, Vercel instantaites your program, sends the exported function the request, gets the function result, send the result to you as an HTTP response, then destroys the program. DOs are exact same thing, but stateful. You upload a JavaScript program to CloudFlare. The external interface is a *class*, not a function. DOs are reachable at a URL. You send a request to the URL and CloudFlare invokes your program, instantiates your class, and sends a particular method on the class the request, gets the result, and send it to you as an HTTP response. So far, just a more complicated version of serverless functions. The difference is that CF keeps that class running for awhile. If another request comes into same URL, CF routes it to the same running program and the same exact JavaScript object in memory. So you can keep state in that JS objects across invocations. That's it, that's the whole pitch. These little javascript programs are low-power (last I looked max ram was maybe 1GB?). But they are cheap to run so you can have zillions of them. This makes them a very useful primitive for certain kinds of problems, like say chatrooms, multiplayer games, etc.
19
18
7
253
50,560
Replying to @ThineAI
@ThineAI knows how I speak and my exact voice and tone than any other AI I use.
7
26
1,895
Nikita BYEr
Ladies and gentlemen, it's time to pass the torch and demote myself to my natural state: a poster. I'll be stepping back from leading product for 𝕏 and will continue on as an advisor. Serving the X community has been the privilege of a lifetime. X is, and will remain, the most important communication technology in history. But running this app is a 24/7 job and it's now time for me to take a breather. The app is seeing unprecedented growth in new users & engagement. We continue to break records every month. We've climbed 70 spots in the App Store since this time last year. And in the last 400 days, we rebuilt almost every aspect of X: the Timeline, the Android app, onboarding, notifications, chat and more. We also launched nearly 30 new products while protecting the integrity of the town square: becoming the first app to show Country-of-Origin on profiles and mounting defenses against AI bots. There's certainly much more work to be done, but our foundation is stronger than ever. None of this would have been possible without the incredible team here. The next leaders will take X to even greater heights with @benjitaylor on design, @singhai on core product engineering and @dinkin_flickaa on mobile engineering -- among many other great people. Thank you to Elon and the X team for welcoming me into the company. See you on the Timeline.
9
11
5
50
12,864
Unfortunately no matter how judicious I am, no matter how much I review the code, it feels impossible to not let the slop slip in. I feel like you can prevent it from overflowing, but without your hands getting dirty, you don't really know the state of the project.
296
148
61
4,391
494,094
We are hiring an SDE Intern for @MerlinAIByFoyer If you are someone who - - Thinks Claude better than Codex - Not heard of Merlin - Haven't seen Silicon Valley - Haven't seen The Social Network - Windows > Mac - Leetcode grinder who can't ship a landing page - Did DSA but never opened a real repo - Has "AI/ML/Tech Enthusiast" in bio - Cant fix a messed up git history - Does not live in terminal then please don't apply, for the rest of you form link in comments.
38
14
5
164
20,116
It's not an Anthropic echo chamber. Some very smart people concentrated in one corner vs the dumbest room in the world, i.e., X, I would happily remain in the corner.
be andrej > join anthropic > shill claude slack bot > say it's a new paradigm > get pushback > "twitter is toxic" > hide in anthropic slack echo chamber > repeat? this is exactly why anthropic is the way they are any pushback from outside the cult is labeled "toxic"
1
309
AFAIR I heard something similar long back - a little search led me to this "Fancy algorithms are slow when n is small, and n is usually small." - one of Rob Pike's 5 rules of programming ..
1
2
1
7
621