Provide in-depth research reports and independent analysis, leveraging data, technology, and economic insights to deliver a comprehensive examination of the blockchain ecosystem, project potential, and market trends. REX Launches a 2x Leveraged ETF Linked to Bitcoin-Treasury Company Strive REX Launches a 2x Leveraged ETF Linked to Bitcoin-Treasury Company Strive REX Shares and Tuttle Capital Management have launched a leveraged ETF that aims to deliver twice the daily return of Strive, a publicly traded Bitcoin treasury company and asset manager. The T-REX 2X Long ASST Daily Target ETF (ticker: ASSX) began trading on Cboe. It seeks 200% of the daily price movement of Strive's stock, before fees and expenses. The fund resets its leverage daily, so returns over longer periods may differ significantly from twice Strive's performance. Unlike spot Bitcoin ETFs, ASSX does not hold Bitcoin or track its price; it provides leveraged exposure to Strive's stock. The companies also offer similar 2x leveraged ETFs linked to other crypto-related firms. Strive currently holds 25,000 Bitcoin, ranking it as the fifth-largest public corporate holder. The company recently funded a purchase of 469 BTC through the sale of its perpetual preferred shares. On Friday, Strive's stock rose 6.4% to close at $30.09. cryptonews.ru 10h ago cryptonews.ru 10h ago NEAR Demonstrates Strongest Growth Since Early February 2025 NEAR Demonstrates Strongest Growth Since Early February 2025 NEAR token surged over 26% in 24 hours, nearing $3.80, its highest level since early February 2025. This rally is attributed to the broader crypto market recovery and the launch of perpetual futures trading via Hyperliquid, allowing direct trading from user accounts. Analysts like Michael van de Poppe suggest the price could target $5, noting the chart's strong performance, while others advise buying potential dips. However, technical indicators warn of a possible correction. The NEAR Relative Strength Index (RSI) reached 77, entering overbought territory. Furthermore, a significant net inflow of NEAR tokens to centralized exchanges suggests increased selling pressure may be building. The rally coincides with ongoing discussions about NEAR's tokenomics, including proposals for a fixed supply and a sovereign development fund, alongside the protocol's shift to a default confidential transaction mode. cryptonews.ru Yesterday 20:15 cryptonews.ru Yesterday 20:15 Bitcoin Fluctuates $1,500 as Rate Hike Lands, 16 Officials Eye Another Increase Bitcoin Fluctuates $1,500 as Rate Hike Lands, 16 Officials Eye Another Increase After the Fed raised interest rates by 25 basis points for the first time in three years, Bitcoin fluctuated violently between $75,000 and $76,500. Sixteen officials forecast at least one more rate hike within the year, but analysts believe Bitcoin may be more resilient to the current interest rate shock than stocks. 2 days ago 09:07 2 days ago 09:07 ZEC Surpasses $1,250 and Nears $1,500: A Technical Frenzy or the Eve of a Pullback? ZEC Surpasses $1,250 and Nears $1,500: A Technical Frenzy or the Eve of a Pullback? ZEC swiftly rose from around $1,110, breaking through a key pivot point. Market attention has now shifted to the two major resistance levels at $1,375 and $1,500. While short-term indicators have strengthened, achieving the higher targets still hinges on the ability to firmly maintain its position above these critical price levels. 2 days ago 07:24 2 days ago 07:24 Shocking, Has Google Cracked RSI? AI Learns to Evolve Infinitely Through 'Dreaming' Shocking, Has Google Cracked RSI? AI Learns to Evolve Infinitely Through 'Dreaming' Google and collaborators have published a groundbreaking paper on "Dream-RSI," a novel method for Recursive Self-Improvement (RSI) in AI systems. The core idea is that an AI agent can recursively improve its own exploration strategy by "dreaming"—replaying and learning from its past exploration history stored in a structured tree. The framework separates a heavy "discovery agent" (powered by models like Gemini) from a lightweight "exploration strategy" (Python code). The agent's exploration results are saved in a "discovery tree." To improve, a "strategy developer agent" rewrites the exploration code, and new candidate strategies are evaluated not by costly real-world reruns, but by fast "dream" simulations replaying the historical tree. Only strategies scoring higher (balancing result quality, compute cost, and parallelism) are deployed, ensuring non-degrading performance. In tests, Dream-RSI dramatically outperformed baselines. For a Lasso solver task, it achieved comparable results in 317 LLM calls vs. 51,200 generations for a standard evolutionary search. In GPU kernel optimization, it matched baseline performance with 2.43x fewer generations or achieved significantly better results with the same budget. Notably, injecting high-level human guidance ("summarized experience") into the agent hurt performance, as it constrained the search space. Dream-RSI's code-level optimization on concrete history proved superior. This work suggests a new RSI paradigm where AI improves not by modifying its core model weights, but by autonomously optimizing the surrounding control code, using history as a simulated world for rapid, low-cost learning. This approach could be a significant step towards more advanced AI self-improvement. marsbit 09/17 04:47 marsbit 09/17 04:47 Balancer Eyes Wind-Down as V3 Revenue Fails to Replace V2 Income Balancer Eyes Wind-Down as V3 Revenue Fails to Replace V2 Income Balancer, a decentralized finance protocol, is considering a graceful shutdown after its restructuring efforts failed to boost revenue following a $128 million hack. Despite reducing costs and launching new products after closing Balancer Labs in March, revenue remains heavily reliant on older V2 pools, with the new V3 pools failing to generate sufficient income. CEO Marcus Hardt acknowledged fulfilling product launch promises but not creating revenue. A governance proposal posted Monday outlines a wind-down plan. Balancer's treasury, holding over $9 million, would distribute funds pro-rata to BAL token holders. Revenue has plummeted since a November 2025 exploit on the V2 platform, falling from $1.13 million in October 2025 to just $56,781 by August 2026. The plan involves ceasing business development next month, giving liquidity providers until October 30 to withdraw funds. Protocol fees will drop to 0%, and only withdrawal infrastructure will remain operational from November 1, with the DAO shutting down. The wind-down is estimated to cost $400,000. Hardt warns delaying would further deplete treasury resources without benefit. BAL token holders will vote on the proposal via snapshot from September 25-29. If approved, the first fund disbursement is expected in May 2027, requiring token burning. Subsequent disbursements will handle remaining funds. If rejected, Balancer will continue under its current structure. TheNewsCrypto 09/15 09:04 TheNewsCrypto 09/15 09:04 CryptoQuant: Bitcoin Needs a Breakout Above $81,700 CryptoQuant: Bitcoin Needs a Breakout Above $81,700 CryptoQuant states that Bitcoin needs to break and hold above $81,700 to confirm a new bull market. Despite a 24% gain over two weeks, BTC price has stalled, trading around $77,100. Analysts identify key resistance levels: the strongest near-term zone is $77,100-$80,200, followed by the 365-day moving average near $81,700—a historical benchmark for bull cycle starts. Further resistance lies at $83,600 and $88,700. Support is seen at the 200-day MA around $70,000 and a significant demand zone of $62,000-$65,000. While the overall outlook remains bullish, the CryptoQuant Bull Score Index has dropped from 80 to 60, indicating a slight deterioration; sustaining momentum requires absorbing sell-side pressure and boosting demand. Other analysts note BTC is currently squeezed between large buying walls from long-term holders at $84,000-$87,000 and short-term holders at $62,000-$65,000. cryptonews.ru 09/13 16:30 cryptonews.ru 09/13 16:30

25 Fields Medal Laureates Warn: The AI Race Is Harming Mathematics 25 Fields Medal Laureates Warn: The AI Race Is Harming Mathematics On September 11, 2026, 25 Fields Medal laureates issued an open statement titled "The Deep Mismatch Between AI and the Aims of Mathematics." They warn that the tech industry's race to use unsolved mathematical problems as benchmarks for AI performance is harming mathematics itself. The mathematicians argue that the core aim of mathematics is conceptual understanding, not merely generating true/false statements. The rapid, mass production of solutions by AI, often published hastily without proper methodology, attribution, or integration into mathematical canon, risks destroying the intellectual soil where new ideas grow. It threatens the human chain of understanding and the development of the ability to ask new questions. The immediate trigger was OpenAI's September 8 announcement. Using 10,000 AI agents over 88 hours, its internal system reportedly found a proof that Navier-Stokes equations can develop a finite-time singularity. While OpenAI stated it is not claiming the Clay Millennium Prize, similar AI-assisted results on related equations were announced by mathematicians from NYU and Anthropic. The signatories, a historic group including Artur Avila, Terence Tao, Maryna Viazovska, and Peter Scholze, acknowledge AI's potential to accelerate genuine understanding. However, they stress that whether the impact is beneficial or destructive depends on human decisions controlling the technology. They call for urgent discussion within the mathematical community and with AI developers. The conflict highlights a widening gap between the speed of AI-generated results and the traditional, slower pace of scientific verification and knowledge integration. In the broader context of the corporate race toward Artificial General Intelligence (AGI), mathematical breakthroughs risk becoming showcases for investors rather than contributions to enduring human knowledge. cryptonews.ru 09/12 21:16 cryptonews.ru 09/12 21:16 After the $320 Million Liquid Network Incident: When One Line of Defense Fails, What Can Digital Asset Platforms Still Safeguard? After the $320 Million Liquid Network Incident: When One Line of Defense Fails, What Can Digital Asset Platforms Still Safeguard? Following a series of recent security incidents in the digital asset space, a key question resurfaces: how should platforms define true security? This is highlighted by the $320 million exploit on the Bitcoin Liquid Network, where assets were transferred not due to compromised keys, but a software vulnerability. Similar events on Cosmos EVM and at Triple-A, involving code bugs and social engineering, underscore a common theme: breaches are often inevitable, but their ultimate impact depends on the depth of a platform's defenses. The critical lesson is that security shouldn't rely on a single, unbreachable gate. The real measure is what happens after the first line fails. Does one compromised credential grant full system access? Can a single point of failure lead to massive asset loss? Effective security lies in layered protections—isolating identities, permissions, critical operations, and asset storage so that a breach in one area is contained. Examining BIT's (formerly Matrixport) updated Trust Whitepaper reveals this philosophy in practice. Its system employs principles like least privilege access, multi-party authorization for sensitive actions, continuous behavioral monitoring, and keeping most assets in cold storage. This creates a series of "gates": a stolen identity doesn't grant full permissions, having a permission doesn't allow independent action, and an online breach doesn't expose core assets. Furthermore, BIT grants its security team a "veto power" to halt projects with unacceptable risks, ensuring security assessments directly influence business decisions. As platforms expand into diverse assets like stocks and RWAs, trust also requires verifiable structures—clarifying which regulated entities handle assets, their roles, and the applicable safeguards. Ultimately, while preventing all attacks may be impossible, the goal is to build resilient systems where a single failure cannot cascade. True security and trust are demonstrated not by claiming imperviousness, but by constructing and maintaining verifiable, interconnected defenses that limit any breach's scope. marsbit 09/11 08:21 marsbit 09/11 08:21 IonQ Named the Required Quantum Computer Power to Attack Bitcoin IonQ Named the Required Quantum Computer Power to Attack Bitcoin IonQ has estimated the resources required for a fault-tolerant quantum computer to attack Bitcoin's cryptography. According to their research, targeting the secp256k1 elliptic curve used in Bitcoin would require approximately 20,000 physical qubits (specifically 19,397), 1,457 logical qubits, and around 39 million logical Toffoli gates. With this configuration, running Shor's algorithm to compute a discrete logarithm would take about 26 days per attempt. This is presented as the first complete, architecture-specific resource estimate, factoring in the compiler, hardware architecture, and error correction for IonQ's trapped-ion "Walking Cat" design. The company links these requirements to its roadmap, anticipating such systems could be built around 2028. The threat is specifically to the Elliptic Curve Digital Signature Algorithm (ECDSA) used for authentication, not to Bitcoin's mining or consensus mechanism. However, IonQ notes that post-quantum signature algorithms like ML-DSA and SLH-DSA are not vulnerable to this class of attack. Simultaneously, IonQ announced an $8.18 million contract with Congruity360 to build quantum-resistant network infrastructure for U.S. corporate clients in sectors like finance and healthcare, integrating IonQ's post-quantum cryptography and quantum key distribution devices. cryptonews.ru 09/08 17:31 cryptonews.ru 09/08 17:31 市场分析 8,336 news items 项目动态 6,435 news items 监管政策 5,912 news items Политика регулирования 1,509 news items Analyse de marché 3,945 news items AI 3,463 news items Prediction 443 news items Cross-Domain 2 news items World Models 12 news items Foundation Models 2 news items Trading Strategies 1 Turkey detains 175 people in $266 million crypto and forex scheme case 2 Coinbase CEO Armstrong Pushes Back as WSJ Prepares Story Blaming Him for CLARITY Act Failure 3 Not Opposing AI, Just Opposing the Idea That Everything Must Be AI 4 Claude Writes 80% of the Code, Nearly Crashes Anthropic, CI Surges 25x in Six Months 5 Revolut Says No Direct Contact After Public $3M Ransom Demand Market Analysis 1 Anthropic Establishes a Biology Laboratory, Not Specifically for Drug Discovery 2 Yesterday's Press Conference's 'Three Words' and 'One Response': Wall Street Puzzling Over 'Wash's Approach' 3 Everything in Bitcoin Depends on This Level: CoinShares and Glassnode Point to the Same Area, Explaining Two Events That Could Trigger a Rally! 4 Crypto Gets a Percentage Quota 5 Blockchain Capital: Lessons from the Crypto Market, Worth Re-reading for AI Investors