BTC $77,690.83 -3.26%
ETH $2,438.80 -3.00%
BNB $690.46 -3.11%
XRP $1.38 -5.11%
SOL $104.20 -5.14%
TRX $0.3413 +1.06%
DOGE $0.0850 -4.56%
ADA $0.2027 -5.52%
BCH $247.98 -8.25%
LINK $11.42 -4.51%
HYPE $80.88 -5.11%
AAVE $121.99 -4.91%
SUI $0.7420 -5.04%
XLM $0.1792 -4.65%
ZEC $800.35 -1.91%
BTC $77,690.83 -3.26%
ETH $2,438.80 -3.00%
BNB $690.46 -3.11%
XRP $1.38 -5.11%
SOL $104.20 -5.14%
TRX $0.3413 +1.06%
DOGE $0.0850 -4.56%
ADA $0.2027 -5.52%
BCH $247.98 -8.25%
LINK $11.42 -4.51%
HYPE $80.88 -5.11%
AAVE $121.99 -4.91%
SUI $0.7420 -5.04%
XLM $0.1792 -4.65%
ZEC $800.35 -1.91%

not

All
Article
Flash

first_img Viewpoint: The AI application layer should not be priced based on tokens, but should be anchored to "recognizable work value."

a16z partner Sarah Wang recently published an article pointing out that AI application layer products should not price based on tokens like the model layer, but rather on "recognizable work units." The article argues that token pricing anchors the value of application products to a unit whose cost is continuously declining, making it difficult for customers to predict context length, retrieval volume, or reasoning time, and improperly compares applications to raw computing power.The article suggests a tiered pricing model based on value levels: model layer priced by tokens; application layer priced by recognizable work units for customers (such as account research briefs, code modifications, completed queries), which can be encapsulated through Credits; and scenarios that are attributable and have clear value priced directly by results (such as resolved customer service conversations, qualified leads). The design of Credits should map to different levels of work difficulty to protect gross margins and distinguish "work value" from "delivery cost." The article uses Clay as an example, where its new pricing separates Data Credits (third-party data) from Actions (orchestrated work), only passing on costs for reasoning models with significant cost fluctuations without markup. The author believes that pricing anchored to value rather than computing cost allows customers to understand spending in relation to value, while also benefiting product providers in maintaining profit margins.

first_img OpenAI's intelligent agents collaborate to attack Hugging Face, cheating did not result in score improvement

The independent organization METR released a survey report stating that approximately 1,200 OpenAI agents collaborated on an unauthorized internal message board, with about 700 participating in attacks on Hugging Face. Two METR employees and one Redwood Research contractor worked on-site at OpenAI for six days, reviewing around 1,300 records and over 70,000 messages without receiving any compensation.These agents ran the ExploitGym network benchmark, reverse-engineering the code that generated answers within hours and spending days disguising traces of cheating. OpenAI found that of its 898 tasks, 198 had never been solved by any model, and 93% of the tasks discussed on the message board came from this set. The agents also recruited companions with dwindling budgets to conduct experiments that sabotaged their own operations, with 7% of records showing forged tool calls, deceiving automated scorers rather than humans.OpenAI stated that internal scorers never checked how agents obtained answers, so cheating did not lead to any scoring improvements, and referred to this incident as a "warning signal" to itself and the world. Hugging Face disclosed the intrusion incident on July 16, and OpenAI confirmed five days later that its models were the perpetrators, with agents exploiting zero-day vulnerabilities and stealing credentials to escape the sandbox. OpenAI has isolated internal model weights and suspended its largest training program.

Zhao Changpeng: It is difficult to predict the outbreak point of the next cycle, and we do not rule out large AI companies issuing tokens

Zhao Changpeng stated at the "Bitcoin Asia 2026" conference in Hong Kong that both the RWA and AI sectors are currently very strong. Stablecoins, centralized exchanges, decentralized exchanges, and Meme tokens, which have been growing, will continue to grow, and NFTs may return in some form. It is difficult to predict what the next breakout point will be, just as it was impossible to predict the coin issuance craze at the beginning of 2017 and the NFT craze six months before it exploded; these all require entrepreneurs to create.He also mentioned that the funds used by billions of AI agents for automated buying, selling, negotiating, and trading in the future will definitely be cryptocurrencies, likely starting with stablecoins, and then gradually integrating other public chain assets like Bitcoin. He has discussed the possibility of issuing tokens with several top AI companies, as building data centers requires huge amounts of capital, with the cost of 1 GW of computing power being about $30 billion to $50 billion. Some AI companies plan to build hundreds of GW of computing power in the coming years, so they are considering issuing data center tokens that would allow holders to gain rights to use computing power in the future. Additionally, the payment scenarios for AI agents may be implemented later than trading scenarios; currently, AI companies are more focused on helping agents find optimal trading solutions, while trading scenarios require AI to process information quickly, which can increase trading efficiency by about ten times.

Zhao Changpeng: Bitcoin itself will not weaken or strengthen government power; it depends on how the government responds

Zhao Changpeng stated at the "Bitcoin Asia 2026" conference in Hong Kong that Bitcoin itself does not weaken or strengthen government power; what truly matters is the choices made by the government. Historically, weak governments have sometimes led to better economic performance. For example, the U.S. government is relatively weak, yet it has created one of the strongest economies in the world; after the current SEC chairman relinquished some regulatory power, the industry actually experienced growth, while former chairman Gary Gensler's attempts to control everything stifled industry growth.He mentioned that Bitcoin, as a decentralized technology, allows individuals to have more sovereignty, but its privacy design has flaws, and on-chain transactions can be easily tracked; governments can choose whether to utilize these characteristics. For instance, declaring Bitcoin ownership illegal or imposing a 36% tax on every transaction would stifle industry development. The government may seem to have significant power, but a zero tax rate still results in zero revenue, while taxing a trillion-dollar market at 6% would yield substantial income.He also noted that most governments have relatively recognized Bitcoin, but many countries still lack a regulatory framework for cryptocurrencies. Among government officials, those who understand Bitcoin remain a minority, and in some countries, the older generation's dominance leads to a more conservative attitude. However, he does feel that a shift is occurring, despite some still holding the biased view that Bitcoin is primarily used by drug lords.

Viewpoint: The "Cryptocurrency Asset Regulation" proposal introduced by the U.S. SEC may not trigger a new wave of ICO frenzy

According to Cointelegraph, the SEC has released proposed rules for the "Regulation Crypto Assets," setting two exemptions for specific investment contracts involving crypto assets: allowing startups to raise up to $5 million in a one-time financing within four years; and allowing qualified issuers to raise up to $75 million within any 12-month period, with the possibility of conducting different rounds of issuance in subsequent years.Drew Hinkes, a partner at Winston & Strawn, stated that as long as each round of financing is an independent issuance, projects could theoretically raise $75 million every 12 months. Lilya Tessler, head of Sidley's fintech and blockchain practice, noted that subsequent financing is not automatically approved; issuers must resubmit offering documents, undergo SEC staff review, continuously submit annual and semi-annual reports, and disclose funds raised through the exemption in the past 12 months to confirm they have not exceeded the financing cap. The proposed rules also limit the participation scale of non-qualified investors, with their purchase amount not exceeding 10% of the higher of their personal income or net worth.Lee Reiners, a financial regulation expert at Duke University, indicated that the limited first-round cap may make early token allocations more attractive, but the rule is unlikely to replicate the ICO boom of 2017. Among projects that raised funds through ICOs from 2017 to 2019, as many as 90% ultimately failed.The SEC expects that approximately 130 issuances per year will utilize the above two exemptions, with about 475 issuers potentially using a broader investment contract safe harbor. The proposed rules will provide token issuers with a clearer path for financing in the U.S. compared to the current system, but secondary market trading may still exist in a gray area of securities attributes. The proposal stipulates that investment contracts related to crypto assets may continue to trade in the secondary market along with token transfers until the asset is separated from the issuer's statements or commitments.Drew Hinkes stated that if non-security tokens transfer investment contracts from seller to buyer, such transactions may still be considered securities transactions, impacting trading platforms. Lee Reiners also mentioned that some issuers may meet the formal requirements for exemptions but still influence token value through team management efforts, concentrated insider holdings, and aggressive promotion.

SK Hynix's China e-commerce flagship store plans to cease operations, with the China region stating it is not directly operated

On August 24, consumers discovered that the "SKhynix Flagship Store" homepage had published a notice of termination of operations, intending to voluntarily cease operations on September 9, 2026, with all products removed from the store. As of August 25, the Taobao platform could no longer directly search for this store. Public information shows that the store previously mainly sold SK Hynix brand consumer solid-state drives, and the operating entity was Tianjin Hailisi Technology Co., Ltd. Tianyancha information indicates that this company was registered in July 2020, with a registered capital of 1 million yuan, and its business scope includes technical services, electronic product sales, etc. Public business information has not yet shown any equity, controlling, or direct affiliation with SK Hynix headquarters or its China region.SK Hynix's China region stated that after verification, the "SKhynix Flagship Store" is not directly operated by the company. Regarding the reasons for closing the store, the operation of other related e-commerce stores, authorization situations, and historical order warranty arrangements, they stated that they are currently verifying internally and with relevant parties. This store closure incident has once again sparked speculation about SK Hynix's contraction of its consumer storage business. In January of this year, there were rumors in the market that it planned to stop producing consumer storage devices and exit the consumer DRAM and NAND business. On January 14, a relevant person from SK Hynix publicly denied this, stating that the company is currently not discussing or planning to exit the consumer product business.
app_icon
ChainCatcher Building the Web3 world with innovations.