engine

Сообщение от ChainCatcher, платформа аудита безопасности блокчейна Sherlock сегодня объявила о запуске Sherlock Audit Engine. Этот движок интегрирует самые мощные методы AI безопасности в совместный процесс проверки, направленный на открытие новой эры нативного AI аудита безопасности.Sherlock заявляет, что команда потратила несколько месяцев на разработку этого движка в закрытой среде и тестировала его на коде с высоким уровнем риска. Audit Engine теперь официально запущен и может проводить координированные AI проверки безопасности для кода, такого как смарт-контракты.

Сообщение ChainCatcher, согласно эксклюзивному отчету Titan Media, 30 июля ByteDance выпустила внутреннее письмо, объявив о слиянии команды продукта Feishu с командой продукта Doubao, образовав новую команду продукта Doubao, которой будет единолично руководить глава Doubao Чжао Ци, а глава Feishu Се Син будет отчитываться перед Чжао Ци. В системе GTM команда Feishu GTM будет интегрирована с командой Volcano Engine, образовав новую организацию ToB GTM "Платформа услуг креативности (Creativity Service Platform)", которая будет полностью отвечать за маркетинг, продажи и обслуживание клиентов облачных услуг ByteDance, таких как MaaS и SaaS, под руководством главы Volcano Engine Тан Дай, а глава продаж Feishu Линь Чань и глава стратегии и маркетинга Feishu Ши Чжи Цзюнь будут отчитываться перед Тан Дай.Это слияние является крупнейшей корректировкой ToB-бизнеса со времени реализации сегментации бизнеса ByteDance в 2021 году. После завершения реорганизации продукты и услуги Feishu останутся неизменными, Feishu будет более глубоко сотрудничать с Doubao в области производительности. Корпоративная версия Doubao, в разработке которой активно участвовала команда продукта Feishu, уже запущена в бета-тестирование среди некоторых клиентов Feishu. В настоящее время Feishu охватывает более 10 миллионов корпоративных пользователей, и это слияние further улучшит интегрированную координацию ToB-бизнеса.

In mid-2026, multiple data points indicate that AI tools are generating significant hidden costs. A survey by Entelligence AI of 2,444 companies found that for every $1 spent on AI, $0.44 is used to fix vulnerabilities, $0.27 is spent on rewriting AI-generated code, and $0.11 is consumed by audit and merge delays.The 2026 report from Lightrun further notes that 43% of AI-generated code still requires manual debugging in production environments after passing quality checks, and no surveyed engineering leaders expressed complete trust in the AI outputs that have been deployed.On the infrastructure front, Oracle has accumulated approximately $108 billion in total debt and is raising another $50 billion in 2026 for AI data center construction, with free cash flow nearing negative $13 billion. Over $300 billion of its $553 billion backlog is concentrated in OpenAI, which incurred a loss of about $14 billion last year. In terms of talent, OKX CEO Stax Xu stated that AI agents have exposed employees who rely on impression management rather than outcome production while accelerating execution. The exchange has now incorporated AI proficiency into its employee evaluation system.

The prosecution materials released by the U.S. Attorney's Office for the Southern District of New York show that Google security engineer Michele Spagnuolo has been arrested and charged for allegedly using significant non-public information to bet on Google user search result-related markets on Polymarket.The prosecution materials state that Spagnuolo used Google internal tools to track the list of the most searched people in 2025 and transferred approximately 3.8 million USDC to a Polymarket address; his associated account AlphaRaccoon placed a bet that D4vd would become one of the most searched people in late November after accessing internal tools for several hours.The prosecution materials also state that AlphaRaccoon transferred 5 million USDC from the Polymarket account to a wallet and moved the funds through exchange services and privacy tools, with some of the funds ultimately entering an account at a payment processing institution in Italy. Spagnuolo is accused of profiting over 1.2 million dollars from the related transactions and currently faces charges of commodity fraud, telecommunications fraud, and money laundering.

The Ethereum Foundation stated that the Open Intents Framework aims to serve as a shared infrastructure for cross-chain intents, providing a modular open framework for deploying, discovering, and executing intents for the ecosystem.At the same time, LI FI announced the launch of LI FI Intents, a full-stack execution engine for same-chain and cross-chain transfers, which competes for quotes and completes order execution through a professional Solver network. LI FI documentation shows that its Intent/Solver Marketplace is one of the official foundations of the Open Intents Framework, and the Ethereum Foundation claims that the framework is entering a phase of large-scale adoption.

"On-chain detective" ZachXBT stated that the hacker "Dritan Kapplani Jr" transferred approximately $2.59 million in assets, including 1.99 million DAI and 259 ETH. The related funds were transferred from address 0x4487...bba6 to address 0x67ec...125d, and the stolen funds are currently in a stagnant state.ZachXBT indicated that it released an investigation report on May 12, detailing the connections between Dritan Kapplani Jr and Trenton (Trent) Johnson in a social engineering theft case involving 185 bitcoins (approximately $13 million).

The cryptocurrency research institution Delphi Digital released the latest report "How Far Can Saylor Stretch It," which systematically analyzes the Bitcoin (BTC) funding expansion mechanism of Strategy, pointing out that its financing structure is transitioning from "low-cost accumulation" to the "diminishing marginal efficiency" stage. The report shows that in the current asset accumulation system centered around Bitcoin, STRC has become the core financing tool for Strategy's continuous purchase of BTC. Initially, it relied on a significant premium in MSTR's stock price (mNAV far exceeding BTC's net value) to achieve a positive cycle of "issuance leads to accumulation," but as the valuation has fallen back to about 1.24 times the EV-based mNAV, the BTC per share enhancement effect from common stock issuance is nearing breakeven.At the same time, while convertible bond tools have played an important role historically, they have accumulated about $8.2 billion in principal and will face concentrated repayment pressure after September 2027, putting long-term sustainability of the financing structure under pressure. STRC provides a continuous financing source for Strategy by offering approximately 11.5% annualized monthly dividends to income-oriented investors, to maintain the pace of BTC purchases. However, this mechanism also introduces ongoing cash flow obligations, meaning that each round of financing increases BTC assets while simultaneously accumulating future dividend burdens.The report emphasizes key risk scenarios: if BTC prices remain stagnant and MSTR's premium fails to recover, then the "STRC financing purchase gain" may be gradually offset by "common stock dilution and dividend obligations." Although the company's approximately $2.25 billion cash reserves can cover about $1 billion in redemption pressure in 2027, larger-scale debt and dividend structures in 2028 still need to be addressed. Additionally, the current authorized issuance limit of about $28.3 billion for STRC becomes a critical constraint point. Once the limit is reached, the ability to purchase new BTC may slow down, but existing dividend obligations will continue to exist, thus altering the overall BTC per share dynamic growth path.

The BAI platform announced user data for May 11, showing that the platform gained 8,756 new users that day. While maintaining strong growth, the proportion of Stripe payments among the core paying group has risen to 69.0%, indicating a steady improvement in retention rates among traditional developers and production-level users. In terms of business structure, the DeepSeek-V4 series has contributed nearly 60% of Token consumption due to its extreme cost advantages. This series of models, with ultra-low latency, high concurrency, and high throughput capabilities, is becoming an indispensable core execution engine in the AI Agent era. The B.AI platform will continue to provide developers with efficient and economical model services, helping more AI-native applications to be implemented.

According to CoinDesk, Ripple announced on Monday that it will share its internal intelligence on North Korean hackers with the cryptocurrency industry threat intelligence sharing organization Crypto ISAC to help businesses identify coordinated infiltration actions.This move comes against the backdrop of a recent shift in attack patterns faced by the cryptocurrency industry. Ripple stated, "The strongest crypto security posture is a shared posture. A threat actor who has not passed a background check at one company may submit resumes to three other companies in the same week. Without shared intelligence, each company starts from scratch."

North Korean hackers had months of offline contact and infiltration with employees of Drift Protocol before launching one of the largest social engineering attacks in the history of the cryptocurrency industry, resulting in a loss of approximately $285 million.Data from blockchain security company TRM Labs shows that North Korean hackers have accounted for 76% of all cryptocurrency hacking losses in 2026 to date.

Coinbase stated that it is optimizing the rule creation process in its anti-fraud system by integrating machine learning models with a rules engine, achieving more efficient risk management. It also proposed a dual-track strategy of "models responsible for long-term defense, rules responsible for rapid response," and built a unified framework to create a feedback loop between the two: rules are used to capture new types of fraud and train the model in reverse, thereby continuously enhancing overall defense capabilities.In terms of specific optimizations, Coinbase has transformed the previously manual rule creation process into a data-driven and automated recommendation system by restructuring data, automating schema evolution, and introducing notebook-based analytical tools, significantly improving efficiency. Among these improvements, the performance of rule backtesting has increased by more than 10 times, and the overall response time has been reduced from several days to a few hours. Additionally, the new system uses machine learning to recommend parameters, helping to reduce false positive rates while combating fraud and minimizing the impact on normal users. Coinbase indicated that the next step will be to advance event-driven automatic rule generation and explore the "one-click conversion" of efficient rules into model features, further moving towards an automated risk management system.

According to The Block, the Ethereum Name Service (ENS) gateway eth.limo experienced a DNS hijacking attack last Friday evening. The project released a post-incident analysis report on Saturday, tracing the attack back to a social engineering attack against the domain registrar EasyDNS.eth.limo stated that DNSSEC validation rejected the attackers' domain server changes, and no users have been found to be actually affected. In response to the recent breaches at the DNS level affecting crypto frontends, eth.limo plans to migrate to EasyDNS's stricter security services, which no longer support account recovery features.EasyDNS CEO Mark Jeftovic has publicly acknowledged the vulnerability, stating that this is the first successful social engineering attack against customers in its 28-year history.

At the "Crypto 2026: From Cryptocurrency to Smart Economy" themed forum held in Hong Kong, guests including Sign partner Sarah, HashKey Chain Senior Business Development YuYi, Monad Greater China Ecosystem Head Harvey Chen, OmenX Chief Marketing Officer Gisele, and SVP Chain Market Head Celest engaged in a roundtable discussion on "The Revaluation of Crypto and the New Cycle."Regarding why they remain firmly invested in Crypto amid the AI boom, Sarah pointed out that the hotter AI gets, the more use cases there are for Crypto assets, and future transactions between AI Agents will rely on digital assets. YuYi believes that the bear market is precisely when the wheat is separated from the chaff, with compliant RWA (such as Hong Kong's first physical silver token) being implemented, speculative assets being eliminated, and the era of on-chain infrastructure arriving. Harvey Chen stated that high-performance public chains are the foundation for supporting large-scale applications, and Monad is committed to this. Gisele noted that Crypto addresses fairness and value transfer, complementing AI's focus on efficiency. Celest emphasized that AI lacks a native value system and increasingly requires Crypto to provide value authentication and decentralized trust.On the changing understanding of Crypto's value, the guests unanimously agreed that the industry is shifting from "issuing tokens for the sake of issuing tokens" to pursuing sustainable business models, with greater focus on capital efficiency, low costs, and 24-hour settlement as practical values. The value of public chains should be assessed based on real business scenarios and user transaction volumes, rather than narratives and financing.Regarding future growth engines and entrepreneurial advice, the guests generally expressed optimism about compliant RWA, AI payments, and sovereign blockchain directions, advising entrepreneurs to focus on real application scenarios, avoid empty concepts, and maintain a long-term construction mindset.

At the "Crypto 2026: From Cryptocurrency to Smart Economy" themed forum held in Hong Kong, Tencent Cloud's industry architect Alan Nie delivered a keynote speech titled "Cloud + AI Dual Engine: Tencent Cloud Empowers New Growth in Web3."Alan Nie pointed out that Web3 enterprises face three major bottlenecks in global infrastructure, R&D efficiency, and business intelligence. Tencent Cloud deeply integrates the "Cloud + AI" dual engine to provide low-latency infrastructure covering the globe. Among them, the Singapore data center is the only cloud provider in the world that offers four availability zones, and the self-developed TDSQL-C database can achieve elastic scaling in seconds, with Redis single-node performance reaching over 300,000.In terms of AI empowerment, Tencent Cloud launched CodeBuddy (AI Pair Programmer) and WorkBuddy (Personal AI Assistant), which can automatically generate code, batch process office documents, organize meeting minutes, and support multi-agent parallel execution of complex tasks. In financial scenarios, the automation research report reproduction cycle has been shortened from three days to half a day; in on-chain data analysis scenarios, storage costs have been reduced to one-tenth of the original, with query responses reaching millisecond levels.Alan Nie stated that Tencent Cloud is committed to using the "Cloud + AI" dual engine to help Web3 enterprises build the next generation of smart economic infrastructure.

According to the threat intelligence released by the SlowMist security team, its threat intelligence system MistEye has received community feedback regarding an active social engineering attack targeting cryptocurrency users.The attackers contact target users under the pretext of project collaboration, luring them to use a counterfeit "Harmony Voice" software (domain: harmony-voice[.]app) for so-called real-time translation, which is actually malware. SlowMist has synchronized the relevant threat intelligence (IOC) to its corporate clients.