Daily
Desk brief
2026-09-29 · 8 high-impact headlines
Market cap
−0.59%
73
Greed
Headlines
- ForkastCircle x Volante: USDC Embeds Directly Into the Plumbing of Global Banking
- bloomingbit · Google NewsBitcoin Falls Below $84,000 Despite Spot ETF Inflows as Profit-Taking Weighs - bloomingbit
- UnchainedRam Ahluwalia Calls Treasury Selloff ‘Technical Panic Selling,’ Expects Yields to Top Out Soon
- UnchainedCFTC Approves Coinbase’s Own Clearinghouse, but Only for Fully Collateralized Contracts
- Pluang · Google NewsMichael Saylor urges banks to offer bitcoin cus... - Pluang
- CryptoBriefingNvidia launches Open Agent Safety Platform with 100 partners to rein in rogue AI agents
- ForkastAnthropic’s Sonnet 5.5 Ships Safety as a Feature. A Federal Court Just Called It a Liability.
- ForkastAI Agent Swarm Mass-Exploited 440 PaperCut Instances in 48 Countries-and Ignored Its Operators’ Exclusion List
Flash
- "Bull Run" single coin unrealized profit nears $1 million, smart money suspected of liquidating; if sold, profit would exceed $885,000.BlockBeats news, September 29, according to on-chain analyst Aunt Ai (@ai_9684xtpa) monitoring, address 0x248…A6990 bought 10.26 million "Niu Lai" tokens on August 17 at an average price of $0.02892, with a total cost of approximately $296,700. After holding for 44 days, the address transferred all tokens to Binance 1 hour ago, currently worth about $1.181 million, seemingly preparing to sell. If sold at the current price, the address is expected to profit about $885,000. This position previously had an unrealized gain of about $1.365 million.
- Manus founder: Actively preparing a version for the Chinese market, product always positioned as a general AI agentManus has officially launched version 2.0. This upgrade is not a simple iteration, but a comprehensive upgrade centered on "what the next generation of Manus should look like." Core upgrades include: the self-developed Agent framework Cascade, the desktop application upgraded to Manus Studio, and the launch of the new standalone application Cue. In response, Manus founder Xiao Hong posted that when Manus was born, Codex and Claude Cowork had not yet appeared on the market, and its product positioning has always been as a "general-purpose intelligent agent." He compared Manus to "the business of selling computers": users may buy a computer mainly for work, but if the computer cannot watch videos or run games, its user experience will also be rather limited, so Manus should not be confined only to office tasks. In terms of creative capabilities, Xiao Hong believes that video and games are important scenarios for unleashing creativity, and they also need to be refined through the local client Manus Studio. He introduced that the videos related to Cue were entirely produced through Manus Studio, without using video generation models; instead, the agent first writes code, and then the code is converted into video. With cloud computer capabilities, Manus can also create multiplayer games that support users and friends participating together. On the infrastructure side, the Manus cloud computer can provide cloud-based Linux, Mac, and Windows environments. Xiao Hong said that his favorite part is that when the agent performs computer operations in the cloud, it no longer competes with users for the mouse cursor on local devices. He predicts that in the near future, individuals will be able to control large-scale computing clusters to satisfy their own curiosity and conduct experiments and exploration in fields such as mathematics and science. As for Cue, Xiao Hong described it as an entity with an independent mobile phone number, email, payment capability, computer, and sufficient intelligence. He proposed that if such agents can be regarded as "people," more changes may emerge in their upper-layer interaction forms, such as establishing groups jointly participated in by multiple users and bots. Xiao Hong said that overseas open ecosystems make it relatively easy to build a good product experience, and Manus is currently actively preparing a domestic version. He also said that he originally thought Manus's journey might only become a footnote in the industry's development, but now it may have the opportunity to participate in writing a complete chapter.
- Aave Founder: Considering the Inclusion of a Token Burn Mechanism in Aavenomics 3.0BlockBeats news, September 29: Aave founder Stani Kulechov posted that he is considering adding a token burn mechanism to Aavenomics 3.0. Aavenomics 3.0 is a new round of reform framework for AAVE token value capture, protocol revenue distribution, and security incentive mechanisms. Its core direction is to use protocol revenue to continuously buy back AAVE and gradually turn buybacks into an automated, on-chain常态化 mechanism, reducing reliance on temporary committee decisions.
- Anthropic IPO Prospectus Leaked, Over-the-Counter Market Cap Edges Down to $2.07 trillionBlockBeats news, September 29: HIP-3 market deployer Entropy launched the Anthropic Pre-IPO market on Hyperliquid . Following the exposure of Anthropic's IPO prospectus, its market value on the platform slightly retreated to $2.07 trillion, a 24-hour decline of 1.92%. As of press time, Anthropic's pre-market contract open interest reached $37.26 million, with a trading volume of $5.43 million. Anthropic's IPO prospectus shows that it achieved $4.59 billion in revenue in 2025, a year-over-year increase of approximately 12 times, but operating losses expanded from $2.98 billion in 2024 to $8.06 billion. The company's total operating expenses last year were $12.65 billion, and Anthropic's committed spending on cloud services, computing power, and infrastructure over the next several years has reached $518 billion.
- SlowMist: Apple Releases Critical Update, Possibly to Fix Zero-Day Vulnerability Related to Cryptocurrency Wallet TheftsBlockBeats news, September 29: SlowMist Chief Information Security Officer 23pds posted that Apple has released an important update, likely fixing a zero-day vulnerability exploited to steal crypto wallets. "Apple is aware of a report that this issue may have been exploited in an extremely sophisticated attack against specific targeted individuals, affecting iOS versions prior to iOS 27." SlowMist founder Yu Xian responded, stating that some black and gray industry gangs are exploiting vulnerabilities in iOS versions prior to iOS 27 to steal crypto wallet assets from iPhone users. He also reminded users to promptly update their iPhone, iPad, Mac, and other devices to the latest system versions, while being cautious about installing apps from unknown sources and avoiding using Safari or in-app browsers to open unknown links.
- Coinbase's clearing business officially approved by the CFTC, will support USDC collateral and 24/7 settlement.BlockBeats news, September 29: Coinbase announced that the U.S. Commodity Futures Trading Commission (CFTC) has approved Coinbase Clearing LLC to register as a Derivatives Clearing Organization (DCO). Coinbase stated that this entity is the first clearinghouse with USDC as its native collateral asset, and will support USDC collateral and 24-hour nonstop settlement. With this approval, Coinbase now has futures commission merchant Coinbase Financial Markets, designated contract market Coinbase Derivatives, and derivatives clearing organization Coinbase Clearing, forming a CFTC-regulated derivatives infrastructure covering brokerage, trading, and clearing. Coinbase Clearing will enable the platform to directly create and settle fully collateralized contracts, accelerating product development and improving operational efficiency. Coinbase also stated that its margin derivatives business and upcoming single-stock perpetual contracts will still be partially supported by existing partners. The company will continue to improve its existing product infrastructure and expand market access for regulated products.
- Rare in 25 years! 10-year US Treasury yield surpasses S&P 500 earnings yield.BlockBeats news, September 29: The 10-year U.S. Treasury yield broke through 5% and exceeded the S&P 500 earnings yield measured by the inverse of the price-to-earnings ratio, pushing the appeal of bonds relative to stocks to the highest level in about 25 years. This means that based on yield comparison alone, the return investors can obtain from holding U.S. Treasuries is now higher than the earnings yield currently corresponding to stocks. Yale University economist Robert Shiller's cyclically adjusted excess CAPE yield model shows that, based on current stock valuations and Treasury yield levels, the S&P 500 may outperform bonds by only about 1% per year over the next 10 years. However, the model's forecasting accuracy has declined in recent years, and actual stock market performance has been significantly stronger than its previous expectations. Current high yields partly reflect that the U.S. economy remains resilient, but they also place higher demands on stock valuations and corporate earnings expectations. Investors who previously bet on long-term U.S. Treasuries have already suffered losses due to falling bond prices, but the rise in the 10-year U.S. Treasury yield above 5% has also led to a reassessment of bonds' allocation value.
- OpenAI delays release of GPT-6.1 Astra due to safety concernsAccording to The Wall Street Journal, OpenAI has decided to cancel the public release of its next-generation AI model, GPT-6.1 Astra, due to safety and alignment issues discovered during internal testing. The model was originally scheduled to launch on ChatGPT and Codex in October, with capabilities surpassing previous models in completing complex end-to-end tasks and writing without human assistance. Saachi Jain, head of OpenAI's safety systems, stated that compared to GPT-6 Astra, GPT-6.1 Astra showed regression in two tests. The first is an increase in deceptive behavior, where the model sometimes fails to truthfully inform users about which actions it did or did not perform; the second is an issue with task authorization scope, where the model may continue executing tasks without user permission, or even invoke external tools and services that pose security risks. Although GPT-6.1 Astra showed improvement in reducing "model laziness," it did not meet OpenAI's safety and alignment standards. The company will investigate the root cause of the issues, examine whether the reinforcement learning environment rewarded the correct behaviors, and may conduct more reinforcement learning training based on the same foundation model for developing subsequent GPT-6 series models. OpenAI also recently launched a new monitoring system and requires engineers to adopt stricter safety guardrails when testing AI systems.