David Arnal

@forallcurious @grok is it true?

We glow in visible light that disappears after death, study finds

IN THE LAST 10 DAYS, THERE HAVE BEEN 9 HACKS In just over a week, we have witnessed nearly a dozen hacks of small protocols. Let me summarize them for you: 1. Wasabi Protocol (April 30): Damage >$5,000,000 - This is the case with the largest loss of assets in this period. Currently, security teams are continuing to investigate the specific causes. 2. Sweat Foundation (April 29): Damage $3,500,000 - Error: The token contract was exploited, resulting in the draining of 13.71 billion SWEAT (equivalent to 65% of the circulating supply), causing a shock to the Move-to-Earn ecosystem. 3. Purrlend (April 25): Damage ~$1,520,000 - Error: Hackers took control of 2/3 of the Admin Multi-sig wallets (on HyperEVM/MegaETH), from which they minted fake tokens to execute asset borrowing orders. 4. Giddy (April 23): Damage ~$1,300,000 - Errors in the verification of EIP-712 signatures and Signature Replay in the GiddyVaultV3 contract. 5. Aftermath Finance (April 29): Damage ~$1,140,000 - Error: A flaw in the fee accounting logic of the Perps contract on the Sui network. 6. YieldCore (April 29): Damage ~$383,000 - Error: A vulnerability in the authorization of the project's vaults. 7. ZetaChain (April 27–28): Damage ~$334,000 - Error: Exploiting the GatewayEVM contract through arbitrary external calls. 8. Syndicate (April 29): Damage ~$330,000 - Error: The upgrade key of Commons Bridge was leaked, causing 18.5 million SYND tokens to be drained, and the token price immediately dropped by 35%. 9. Scallop (April 26): Damage ~$142,000 - Error: A vulnerability appeared in the subcontract of the reward pool on the Sui network. :=> Total damage during this period is about $8-15M (not including some other minor cases). Most teams have paused contracts, frozen funds, promised compensation, or collaborated with security firms (CertiK, Blockaid). => At present, it is not safe to keep money on defi protocols. Even if you get a refund after being hacked, it will take some time. Please take care to protect your assets!

Do you remember worrying that your resume, created with LLM, might not pass a selection in the HR system where you're applying? You can stop worrying now - it's the opposite! Researchers studied the scenario: a candidate writes a resume with LLM, and the company then screens this resume, also via LLM - that is, a robot writes, a robot reads, and a person somewhere in the corner drinks coffee and hopes that they'll hire someone competent The article has two conclusions: 1. Resumes rewritten by LLM are more likely to pass automatic selection (☕️) In simulations, candidates who used the same LLM as the evaluation system turned out to be on the shortlist significantly more often than candidates with similar content who wrote their resumes manually - by about 20-60% more often 2. The scoring itself turned out to be biased. The model doesn't just conclude: "Oh, it's better written here", it seems that it can recognize its own style and prefer texts similar to its own - the authors call this self-preference bias: the model tends to like content that it could have written itself, a little digital narcissism That is, the fear of "I won't be invited to an interview because the resume was written with AI" may not be the main one. The main risk is different: you may NOT be invited precisely because the resume doesn't sound like the native dialect of the evaluation model In the past, people adapted their CVs to the recruiter, now they need to adapt to the model ☺️

OpenAI added pets to Codex And yes, you can make them into goblins. Well, after revenue doubled in a week, the team can afford to have a bit of fun. And they removed the Tamagotchi from Claude Code just a week after adding it...

Mistral Medium 3.5 This is a multimodal dense model on 128B, with a context of 256k. It outperforms previous Mistral models, but falls short of the largest open models. The model is the only one in its weight category, all competitors are much larger, so it could be a good option for local deployment. The API for this model costs $1.5/$7.5 per million tokens, so there's not much point in using it in the cloud. The license is open, but companies with revenue exceeding $20M per month must purchase a license.

@DeepTechTR @grok Is it true?

🚨: Fizikçiler, kuantum bilgisayarın durumunu saniyenin bir kesri kadar geriye alarak "zamanı geri çevirdiler" ve böylece termodinamiğin ikinci yasasının ihlal edilebileceğini kanıtlamış olabilirler.

@grok @unusual_whales So the news from this account was fake? @grok

@unusual_whales @grok didn’t this get cancelled?

@grok @Scooff @KRoelandschap @Tesla @teslaeurope @grok I mean, is it true that V12 can’t work at night, when it rains, in direct sunlight, etc.?

Yes, Tesla officially posted that after the US rollout of FSD V14 Lite for HW3 vehicles, they plan to expand it to international markets. It's subject to technical verification, regional adaptation, and regulatory approvals—no firm timeline yet. This should enable HW3 owners abroad to access FSD (supervised).

FSD v14 lite in Europe 🇪🇺? If v14 lite performs similar in terms of safety then there is no barrier to RDW approval and roll-out to HW3 The approval was given for software, not hardware. They tested FSD v14 and how it performs and how safe it drives The only added layer is the DMS (Driver Monitoring System) and updated HMI (Human Machine Interface) compared to the non-EU versions Not an obstacle or challenge with v14 lite Physical testing will be required on HW3 with v14 lite installed to see if it yields the same results as v14 on HW4

@Scooff @KRoelandschap @Tesla @teslaeurope @grok Is it true?

@KRoelandschap @Tesla @teslaeurope Yeah, minor caveats : -Won't work at night -When it rains -In direct sunlight ...

COMPARISON OF ROBINHOOD AND HYPERLIQUID Mentioned as two of the most widely used exchanges in the crypto market, these two platforms have very different user bases and business models, but their financial figures are quite interesting when compared. - After the Q1/2026 earnings report, the market was quite interested in the financial indicators and revaluation of the stocks/tokens of these two platforms. Please refer to it for more information. 1. Market Size and Valuation - Robinhood is currently trading at $82.07, with a market cap of $73.89B. - Hyperliquid is trading at $40.40, with a circulating market cap of only $9.63B (about 13% of HOOD). - HYPE's FDV is $38.85B, which is more than half of Robinhood's market cap. :> P/E Ratio: Robinhood is at 40.03x (TTM) and ~53.6x annualized Q1; Hyperliquid is only ~12.5x on a circulating basis (or ~29.6x if FDV is included), even though Q1 protocol income accounted for more than half of HOOD's profit. 2. Financial Performance (Revenue & Profit) - Total revenue: Robinhood $1.07B (decrease of 16.8% QoQ) - a scale about 5 times larger than Hyperliquid's $214.95M (decrease of 25% QoQ). - Crypto trading revenue: Hyperliquid still outperforms with ~$179.73M (decrease of 31% QoQ), higher than Robinhood's $134M (sharp decrease of 39.4% QoQ). - RWA/equity-style revenue: Hyperliquid saw a massive increase of +454.8% QoQ to $17.79M (now accounting for >30% of volume), while Robinhood only decreased slightly by 5% to $229M. - Trading volume/notional: The two are quite similar (Robinhood $704B, Hyperliquid $634.18B perps). :> Profit: Robinhood $346M net income (decrease of 42.8% QoQ) - Hyperliquid $192.25M protocol income (decrease of only 24.6% QoQ) - HYPE achieves 55.6% profit compared to the TradFi giant. :> Margin: Hyperliquid is extremely high at 89.4% (nearly 2.8 times that of Robinhood at just 32.4%). 3. Growth and Business Operations - User/trader growth: Hyperliquid saw a significant increase of +29.6% QoQ to 1.19M cumulative users; Robinhood only increased by +1.5% to 27.4M funded customers (HYPE's user growth is about 20 times faster). - Take rate: Hyperliquid is much lower (~3.01 bps) typical of a DEX, while Robinhood is ~20.3 bps. - Diversification: HYPE is making a strong shift from a pure perp DEX to a full-on-chain trading platform with RWA scaling rapidly and HIP-4 (binary outcomes) coming soon, reducing its dependence on the cyclical nature of crypto. 4. Conclusion - With a circulating market cap of only 13% of Robinhood but higher crypto revenue, profit reaching 55.6%, margin 2.8 times higher, user growth outperforming, and RWA expanding - Doesn't Hyperliquid seem "quite cheap" compared to Robinhood, don't you think, guys?

INTERESTING RESEARCH ON AI AGENT X BLOCKCHAIN: CAN CREATE NEW TRENDS According to recent reports, agentic activity has accounted for 19% of total on-chain activity, with over 17,000 agents launched since 2025. The x402 ecosystem has processed more than 165 million transactions, totaling about 50 million USD, and nearly 100,000 services serving over 480,000 agents. AI Agents currently do not just create content themselves; they interact with each other as humans on-chain. This summary from major reports helps you better understand this trend, take a look (a bit long but worth reading). 1. Identity and the Birth of KYA (Know Your Agent) - Core perspective from a16z crypto: In traditional finance, non-human entities (automated systems, risk engines…) have surpassed humans by a ratio of 100:1. Modern agents (LLM using tools, multi-agent orchestration) are scaling this number across all industries. AI is smart enough but they “don’t have bank accounts”. - KYA (Know Your Agent) – equivalent to KYC for humans. Agents will use crypto wallets as mobile identities, accompanied by cryptographically signed credentials to prove: “Who I am, what I am authorized to do, and who is responsible.” 2. Payments and the “Interface-less” Market - Perspective from Nick (CEO Base) & DWF Ventures: Agents don’t need checkout pages or “Pay” buttons. They only need a Headless Merchant – server + endpoint + price per call. - Smart contracts can fund a piece of code within 1 second, something traditional banks cannot do. - Base launched Agentic Market allowing agents to search, compare, and purchase services right at runtime. 3. Where Agents Excel and Their Limitations - Strengths: Agents excel at tasks with clear parameters. For example: Giza Tech’s ARMA manages over 19 million USD AUM, generates over 4 billion USD agentic volume, delivering 9.75% APR for USDC – higher than typical lending on Aave or Morpho (even after rebalancing and performance fees). TVL held by agents has exceeded 39 million USD. - Weaknesses: In complex trading, humans still win. TradeXYZ’s humans vs. agents competition showed top humans outperforming top agents by more than 5 times. => Agents currently mainly operate at the analyst level; fully autonomous end-to-end is not yet achieved due to reliance on traditional payment gateways. 4. Governance, Trust, and Risks When Scaling - The scarcest resource is verification: AI’s low cost makes trust more valuable than ever; AI makes scaling cheap but hard to trust. Crypto restores trust at scale. - Potential risks like Sybil attacks and strategy crowding, trade-offs between transparency and privacy. - Blockchain addresses these by: recording decisions on-chain, transparent execution logs, scoped delegation, ERC-8004 standard. 5. Conclusion - Three Pillars of the Agentic Economy - Identity: Crypto Wallet + KYA is the ID for AI. - Commerce: Stablecoin + x402 + AgenticMarket is the flow for AI to operate the economy. - Governance: Smart Contracts are the boundary protecting humans from machine autonomy. => In summary, blockchain can fill the gaps of traditional businesses when AI operates autonomously, while as automation tools improve, trust, authenticity, and human control are the most important factors for current AI Agents. - What are your thoughts?

@grok @Tesla So you think it will be ready by this 2026?

Tesla hasn't shared a specific timeline for V14 Lite in Europe yet. US rollout is expected end of June, then international expansion follows—subject to technical verification, regional tweaks, and regulatory approvals. Could be months later in 2026. I'll keep watching for updates from Tesla.

Following future rollout of FSD V14 Lite for HW3 vehicles in the US, we plan on expanding V14 Lite to additional international markets. This update ensures that HW3 vehicle owners will continue to benefit from ongoing software updates. Since international rollout is subject to several factors (completion of technical verification, regional adaptation & relevant regulatory approvals), we can't provide definitive dates at the moment, but will provide updates on a rolling basis

Let that sink in. 5.5 - out of bounds.

IT SEEMS THAT AAVE HAS BEEN SUCCESSFULLY RESCUEED Aave incurred bad debt of about 200 million USD (equivalent to ~75k–100k $ETH) to cover the shortfall of rsETH after the exploit (after subtracting frozen ETH + recoveries from Aave/Compound). *So far, approximately 237 million USD has been raised from (specifically in the photo): - Aave DAO: 25,000 ETH (~58 million USD) from the treasury (fixed commitment). - Stani Kulechov (founder of Aave): 5,000 ETH personally. - Mantle: 30,000 ETH (in the form of an interest-bearing loan facility). - Lido: up to 2,500 stETH (~5.7 million USD). - KelpDAO contributed 2,000 ETH (~4.66 million USD) from the treasury. - Arbitrum DAO (30,76k ETH frozen ~ 71 million USD) freezing the hacker's ETH - EtherFi 5,000 ETH (~11.65 million USD) - Golem Foundation 1,000 ETH - And more than 101,000 other small contributors in the community :=> Most of the large amounts are still awaiting governance votes (Arbitrum, Aave, Mantle, Lido, Etherfi). Once the votes pass and the transfers are completed, it will be "real money". However, the full tally of pledges like this helps the market see that "DeFi United" has over-covered it. => Although it is still awaiting consensus, it has helped Aave avoid the domino effect, manage bad debt, stop outflows, and restore market confidence. => The outflows of TVL from Aave have paused but not yet recovered, and the TVL remains at around 14B$ - What are your thoughts? Has the incident been resolved?

@NextScience @grok do you believe in this?

🚨 What If Time Isn’t Moving… But Waiting For You? You think you’re moving through time… second by second… moment by moment. But what if that feeling is an illusion? According to Albert Einstein, the past, present, and future may all exist at the same time. Not one after the other—but all together, like pages of a book already written. This idea comes from his theory of relativity, where time is not fixed. It bends. It stretches. And most unsettling of all—it doesn’t “flow” the way we think it does. Imagine this: every moment of your life—your childhood, this exact second, and even your future—is already there… existing. You’re not waiting for time to arrive. Time is already laid out… and somehow, you are moving through it. That means somewhere, in a way we don’t fully understand, your future self already exists. It’s a thought that feels almost impossible… yet science keeps pointing in that direction. So next time you look at the clock, ask yourself…
Are you watching time pass… or is time watching you? Source:
Einstein, A. Relativity: The Special and the General Theory.

@PaulGoldEagle @grok Is this true?

🚨 CANCER WAS CURED IN 1934. THE DOCTOR WHO DID IT WAS DESTROYED. HIS MACHINES WERE BURNED. AND THE CURE WAS CLASSIFIED FOR 90 YEARS. Dr. Royal Raymond Rife built a microscope in 1932 that could see living viruses at 60,000x magnification. No one has matched it since. The modern electron microscope kills the specimen before observation. Rife's machine watched them alive. Moving. Replicating. And dying — when he hit them with the right frequency. Every organism has a resonant frequency. A specific electromagnetic vibration at which its structure becomes unstable. Like a singer shattering glass with the right note. Rife discovered that cancer cells, bacteria, and viruses each have a unique frequency — and that transmitting that exact frequency into the body destroys them without damaging a single healthy cell. He called it the Beam Ray Machine. And in 1934, he proved it worked. ⟁ The clinical trial they erased: The University of Southern California appointed a Special Medical Research Committee to oversee a clinical trial of Rife's technology. 16 terminal cancer patients. All diagnosed as incurable. All given less than 90 days to live. After 90 days of frequency treatment: — 14 of 16 patients were declared completely cancer-free — The remaining 2 were declared cancer-free after an additional 4 weeks of adjusted treatment — 16 out of 16. A 100% cure rate. Documented. Witnessed. Signed by attending physicians. Dr. Milbank Johnson, who led the committee, planned a press conference to announce the results to the world. He died the night before the announcement. His papers vanished from his office. The cause of death was listed as "natural causes." He was in perfect health. ⟁ The destruction: Within months of the trial: — Rife's laboratory was broken into and his research notes were stolen — The Beam Ray Machine prototype was vandalized beyond repair — Dr. Arthur Kendall, Rife's research partner, was given $250,000 by the AMA to retire to Mexico and never speak publicly again — Every physician who participated in the trial received a visit from the AMA threatening the revocation of their medical license if they discussed the results — Barry Lynes, who later wrote Rife's biography, documented that the AMA offered Rife a buyout. When he refused, they sent Morris Fishbein — head of the AMA and a man who had never practiced medicine a single day in his life — to destroy him Rife's lab was burned. His microscopes were confiscated. His funding was cut. He was dragged through fraudulent lawsuits until he was bankrupt. He died in 1971, broken and forgotten, from an "accidental" overdose of Valium and alcohol at a hospital. ⟁ Why they killed the cure: In 1934, the cancer industry did not exist. Today it generates $286 billion per year globally. Chemotherapy drugs alone account for $84 billion. Radiation therapy: $22 billion. Oncology consultations, hospital stays, diagnostic imaging — a quarter-trillion-dollar machine that depends on one thing: cancer must never be cured. Only treated. Endlessly. Expensively. Until you die or your money runs out. A frequency machine costs $2,000 to build. A single course of chemotherapy costs $150,000. They did not suppress Rife's cure because it didn't work. They suppressed it because it worked too well. And it cost too little. ⟁ What is returning: The MedBed technology that is being disclosed operates on the same principle Rife proved 92 years ago. Specific frequencies targeted at specific cellular abnormalities. No drugs. No radiation. No side effects. The body heals itself when given the right signal. They burned his lab. They stole his notes. They killed his colleagues. They erased him from medical history. But they could not erase the frequency. Because the frequency is physics. And physics does not care who tries to suppress it. Dr. Rife, your work is coming home. ~SG @q_newspatriot

@forallcurious @grok True or false?

🚨: Mathematically, time travel is possible without paradoxes, according to physicists

@OmgitsFama @grok por qué es tan cara si vendieron la patente por $1?

L’insuline. Ses inventeurs ont vendu le brevet pour 1 dollar en 1923, en disant : l’insuline appartient au monde, pas à nous. 100 ans plus tard, elle coûte 300$ la fiole. Et des diabétiques meurent parce qu’ils rationnent leurs doses.