-T3ch-

Fully locked in on ARC-AGI 3

@tsotchke maybe you have to go the VC route, maybe bootstrapping isn't ur thing, go to SF, get the funding, and grow ur IP

i suppose i could just go back to academia since my efforts to accelerate geometric computing amongst the general population have been largely ignored (while being absorbed by numerous other laboratories with no credit offered to me). it was fun online while it lasted i guess.

@aave full live Mainnet FORK POC Is ready and ran with no mocks

@aave i found a p0 in Aave v4 and I'm not joking or spamming I cannot report it

Certora has formally verified Aave V4. This process covered the Liquidity Hub and Spoke contracts, along with the underlying mathematical libraries. The verification confirmed that the Solidity implementations of V4's contracts are correct according to Certora's formal rules.

@hrkrshnn are u implying that if u hit a rate limit before u can start using it again u have to wait that much of a stretch of time? that's genuinly horrifying, never.

I don't think 5-hour limits make sense anymore, now that agents can run for longer. There should be one weekly or monthly limit.

Incredible Egyptian goal is disallowed because of a foul far away, then same situation a few minutes later and goal for Argentina not disallowed! No VAR, nothing? FIFA again looks like a corrupt joke, playing favorites for stars.

Fuck Zionism Fuck Messi

Messi the goat but this team is coddled by refs This win was stolen from Egypt I have never seen a game this cheated in my life and and I have been watching soccer for weeks now

VAR真的太肮脏了。 史上最精彩的进球，就这样给摧毁了。

@kimmonismus "for us" -- pffft

Anthropic obvioudsly no longer operate on a classic annual model-release cycle. Major generations may still arrive roughly once a year, but meaningful frontier upgrades now appear every two to four months. Claude 4 launched in May 2025. Opus 4.1 followed roughly ten weeks later. If Fable 5 follows a similar cadence after its June 2026 release, a potential Fable 5.1 would most likely land around August or September. It's crazy to think how good Fable 5 is, and that we'll probably get the next iteration next month. Rumor has it Fable 5.1 is already finished. It's hard to imagine what else is in store for us. Again: we are still so early.

"in store for us" -- retard

@aleabitoreddit anything related to current CPO validation?

@aleabitoreddit @SemiAnalysis reports NVIDIA’s Kyber NVL144 delayed to 2028 due to PCB issues, with the CPO-based NVL576 also likely delayed or low-volume from current CPO challenges and CPO NVSwitch pushed to Feynman leaving Rubin Ultra without a proven large scale-up path. does this mean -

Came across an interesting report from SVRC Research called "State of Robotics 2026", published in April. Which listed: 1. Figure AI 2. Agility Robotics $CCXI 3. Apptronik 4. $TSLA 5. Boston Dynamics 6. Physical Intelligence 7. 1X Technologies 8. $AMZN Robotics 9. Covariant 10. Skild AI As the National Champions of the United States robotics program. "The United States leads the world in where robotics is heading: Fundation models, OpenAI-style scaling laws applied to action, autonomous vehicles. While losing the race on where robotics is shipping today." Then it frames: 1. Rare Earths Exposure: from Neodymium for motors to samarium-cobalt for high-temp applications as a critical vulnerability. 2. Actuator dependency. Series elastic actuators, quasi-direct-drive motors, and precision reducers overwhelmingly sourced from Japan, Germany, and China As one of the main vulnerabilities alongside Manufacturing velocity/data collection cost/regulations. Then their take was: "With at least six well-funded US humanoid companies competing for a market still in early formation, we expect at least two significant consolidation events (acquisition or merger) in 2027". With Logistics / E-commerce (like $AMZN / $FDX) and Automotive from $GM to $FORD as being the immediate top use cases for deployment. I think it's just interesting to see a lot of my points I've been talking about reiterated by research firms. Regardless, I do think it's going to be a major frontier race between the US and China. Agility Robotics (which I own), Tesla, Figure, and Apptronik as leaders representing the USA. Competing against Unitree, AGIbot, Ubtech, and others in China.

@hrkrshnn @mignano from the deep web and enterprises, imo I think project GLASSWING was much more than just giving access to their mythos there was def some type of data exchange from propietary code or software etc not outright src code but watv interesting internal data they have

@mignano On training data: the open web is no longer interesting, the frontier models have large amount of private data that they pay for or collect from their products.

I’m running out of reasons to believe the labs will have moats when it’s all set and done. Models? Unless someone reaches runaway self improvement, doesn’t seem like it. Everyone catches up within months. Compute? That’s a cost, not a moat. Same chips for everyone. Capital? You don’t even need a working model to raise right now. Everyone wants in on AI. Talent? Follows the capital. Training data? The web’s been scraped to death. Everyone’s got the same stuff. Lock-in? Routing makes models swappable. It’s a config change, maybe. And smart routing eliminates even that. Regulation? Good luck once the weights are open. AI is being unbundled.

@hrkrshnn thanks!

@QuantvmH Working on it!

This is false. We compete a lot with our AI tool and have earned nearly a million dollars overall and #1 on the US business leaderboard in 2026. The accurate framing is Tenzai is #1 in Q2 in a narrow category: VDPs. Bugs that don't have a bounty. Very little competition there!

@hrkrshnn shows how bad it is when they breached the Mexican government using a model from the dinosaur age (gpt 4.1 🥀)

Links / events: 1. Mexican government agencies breach: the hacker used used Claude Code + GPT-4.1 against 9 agencies. gambit.security/blog-posts/a-s… 2. Claude used after IT compromise to identify an OT access path. dragos.com/blog/ai-assist… 3. Script kiddie's bad opsec: captured Claude + Codex logs from breaches of at least 14 companies. research.openanalysis.net/claude/codex/h… 4. Claude Code data-theft/extortion campaign: at least 17 orgs targeted. anthropic.com/news/detecting… 5. AI-orchestrated espionage campaign / GTG-1002: about 30 targets across tech, finance, chemical, and government. anthropic.com/news/disruptin…

This is absolutely false. There is public evidence of AI being used in hacking ops, often by script kiddies. I'll post the links in the replies. If you talk to people responsible for defending real infra, they are seeing increased AI use by hacking groups. These are mostly private-company incidents and they don't really talk about it in public. All of this feels similar to coding in January: everyone suddenly realized the models were great at coding even though the capability had been there earlier, and the power users have been seeing it much earlier. Cyber will go through something similar, the insiders know the game has changed on both offense and defense. It'll be obvious to the general public in some time.

@hrkrshnn thats legitimately DIABOLICAL, what the fuck🥀🥀😭😭

Models are getting lobotomized on security. This trend will continue.

dumbass, this person hated Islam and was sympathetic to ur retarded ideologies

@hrkrshnn wow I'm dumb I've never thought of it that way, that saves a fuck ton of time

@QuantvmH You can make a GitHub based on the version that's deployed, this is where a lot of the human work goes, knowing what's actually relevant.

If you're a security researcher and could use a simplified, cheaper version of Apex, our autonomous bug hunter, would you use it? I'm thinking about opening up our product for security researchers. We've earned $1M+ in bounties from this. DM me if you're interested.

@hrkrshnn funny question do u find most of the time that GitHub is up to date to deployed bytecode some programs have their entire in scope just explorer contracts lol

@QuantvmH Connect GitHub or endpoints, click a button, answer some questions -> sit back and relax, wait for bugs.