Anthropic · 1249

The US Federal Trade Commission is urged to investigate AI companies' act of destroying books

Comparatively, the US Federal Trade Commission (FTC) is being urged to investigate the acts of some AI companies obtaining AI training data by buying, scanning, and destroying books. According to an open letter obtained by Axios, more than a dozen civil society organizations are calling on the FTC to use regulatory powers to examine what large AI companies call disruptive new methods of data acquisition. Earlier, the “Washington Post” quoted court documents as reporting that Anthropic had spent millions of dollars to buy books and remove book spines to scan the pages and use them to train Claude; Google, Microsoft, and OpenAI have also faced similar copyright lawsuits. These organizations want the FTC to further determine whether such actions constitute unfair competition practices. They believe that by acquiring and destroying physical books, AI companies may actually be emptying the market's key data resources. In particular, some rare books may disappear permanently as a result, while digital companies hold the last few physical copies. Relevant organizations warn that this practice of hoarding and destroying may increase competitors' data acquisition costs, while cutting off important raw materials that AI startups rely on to train models, thereby further expanding competitive barriers for leading AI companies. However, rather than requiring the FTC to restrict AI model training, they want regulators to focus on reviewing the destruction of existing works and intervene before large AI companies use this to establish a market advantage. According to the open letter, this approach is not simply a data acquisition strategy, but may become another structural means for leading AI companies to build a systemic moat that is difficult to overcome. Currently, the FTC under the Trump administration wants to maintain a relatively friendly regulatory environment for US companies, and on the other hand, it continues to release attention to market competition and the monopoly risk of large technology companies.

1m ago

Anthropic Stuffs Mythos 5 Into Claude Security: Businesses Can Use It, But Can't Get the Model

Twitter News, AI Alerts, and Anthropic upgraded Claude Security's underlying scanning model to Claude Mythos 5 and open beta for all Claude Enterprise customers. Once the enterprise connects to the GitHub repository, Mythos 5 scans the code for vulnerabilities and returns the CWE classification, confidence level, severity level, and recommendations for fixes. Claude Security is already online. It was still using Claude Opus 4.7 when it opened beta in late April. The biggest change this time is the addition of Mythos 5, which previously had strict access restrictions, into existing security products. Mythos 5 not only finds bugs, but also turns them into workable attacks, so it has always been open only to audited agencies. Businesses still can't directly use Mythos 5. The model only runs in the background of the scan, and users get bug reports and suggestions for fixes. Subsequent code changes are made in Claude Code, and the company's existing model is still used, and the patch must be manually approved. Scans will calculate tokens according to the existing plan, and there is no need to purchase Mythos 5 separately. Anthropic is also preparing to integrate Mythos 5 into partner security products and launch a $35 million Claude Defender Advantage Fund to help open source projects find and fix bugs.

8h ago

Review of this week's macro hot topics: the US debt crisis, AI infrastructure, and geopolitical conflicts are the main lines of the market this week

Comparing news, the global market this week focused on US debt pressure, AI capital expansion, and the US-Iran economic game. After the US Treasury expanded the scale of long-term treasury bond repurchases, US bond yields declined briefly, but the market feared that fiscal deficits and debt growth pressure would be difficult to ease through liquidity tools. The US federal government debt surpassed 40 trillion US dollars for the first time. The yield on 30-year US bonds once rose to a high level since 2007, and the global long-term bond market was under pressure simultaneously. The minutes of the Federal Reserve's July meeting show that internal hawkish forces are growing, and there are more than three voting members supporting interest rate hikes. Some officials are concerned that tariffs, energy prices, and AI infrastructure investments could drive up inflation. Meanwhile, Federal Reserve Chairman Walsh suggested that in the future, consideration could be given to reducing the number of annual meetings from 8 to 6. Driven by the weakening dollar and risk aversion, gold broke through the 4,600 US dollars/ounce mark this week and rose for the third week in a row; crude oil was higher, supported by the risk of the Strait of Hormuz and expectations of US sanctions against Iran. Geographically, the US-Iran relationship is shifting to putting pressure on the economy. The US plans to weaken Iran's economy by expanding sanctions and economic isolation, while Iran is studying countermeasures against energy transportation nodes, and the safety of the Strait of Hormuz has become the focus of market attention. In the field of technology, AI infrastructure competition continues to escalate. Nvidia guarantees up to $105 billion for the OpenAI data center project, and Broadcom is also planning an AI financing plan of up to $100 billion. Meanwhile, Anthropic's revenue surpassed OpenAI for the first time, and plans to advance IPOs, further intensifying AI companies' commercialization competition. On the capital market side, Yushu Technology skyrocketed on the first day it landed on the Science and Technology Innovation Board. At one point, its market capitalization exceeded 44 billion yuan, and founder Wang Xingxing's net worth increased dramatically. South Korean semiconductor giant SK Hynix announced a repurchase plan of approximately 40 trillion won, and Samsung is also planning to increase shareholder returns. Furthermore, trade negotiations between the US and Canada ushered in a critical window. The US suspended the imposition of up to 50% tariffs on Canadian goods for three days, and the two sides continued to seek trade agreements. The core logic of the market this week still revolves around three themes: whether US fiscal pressure worsens further, whether AI capital investment is forming a new round of asset bubbles, and whether global geopolitical risks are driving safe-haven assets to continue to rise.

9h ago

Anthropic hires Google chip business executives to speed up the deployment of self-developed chips

According to news, Anthropic has hired Amir Salek (Amir Salek), one of the founders of Google's custom chip project under Alphabet (GOOG.O). As this artificial intelligence lab paves the way to enter the field of self-developed semiconductors, Salek joined as part of its hardware layout. Anthropic said on Friday that Salek will join the AI company's computing team. Salek was previously responsible for Google's Tensor Processor (TPU) business until he left office in 2022, and led the delivery of the first seven generations of TPU chips. Anthropic currently purchases chips from multiple vendors, including Nvidia (NVDA.O), Google, and Amazon (AMZN.O). However, the company has recently released signals that it wants to establish its own internal chip business. San Francisco-based Anthropic has begun recruiting people for this project and posting relevant positions.

10h ago

Anthropic's valuation targets $2 trillion, funding may exceed $100 billion

Comparing the news, according to the “New York Times”, two people familiar with the matter revealed that Anthropic recently stated during discussions with potential investors that the San Francisco-based company may seek to raise more than 100 billion US dollars in an initial public offering. According to people familiar with the matter, the offering could also raise the company's valuation to $2 trillion. If successful, Anthropic, which is only five years old, will set the record for the largest public offering in history, surpassing the record set by Musk's SpaceX in June this year — the latter was valued at $1.77 trillion when it went public and raised $85.7 billion. It would also be a major leap forward for Anthropic. The company reached a valuation of $900 billion in a private funding round this year and submitted a listing application in June.

12h ago

[Comparative Daily News Picks] Anthropic plans to include anti-AI sentiment as the main risk factor in the prospectus; Strategy's stock price hit a two-month high, and STRC returned above $96; Bernstein: Even if the “Clarity Act” is not passed, the SEC and CFTC will speed up rule-making; Dalio: The US debt crisis may break out within three years, and it is recommended to increase gold holdings

Daily AI · Crypto · Macro · Market Highlights, Bitpush helps you set priorities ↓ AI · News [Anthropic plans to include anti-AI sentiment as the main risk factor in the prospectus]. According to CNBC, Anthropic is expected to list the public's negative sentiment about artificial intelligence and data centers as a risk factor in the IPO prospectus to be released in the next few weeks. According to people familiar with the matter, Anthropic recently held a pre-listing “market trial” meeting with bankers and investors. Investors focused on competitive pressure, the impact of open source models on profit margins, and the risks that may be brought about by a slowdown in data center construction. Anthropic is currently valued at close to $1 trillion in the private equity market and is preparing to hit a major IPO. However, as Americans' concerns about AI replacing employment and data center expansion heat up, the related backlash sentiment is becoming a new challenge facing the company's listing. The company has previously achieved an annualized revenue operating rate of more than 65 billion US dollars. [Apple cuts Siri and Vision Pro team positions, and resources shift to AI and new devices] Compared to news, Apple (AAPL.O) is laying off employees from various teams responsible for Siri's digital assistants and Vision Pro headsets. The total impact of this layoff is more than 200 people. Of these, about 100 jobs in the Vision Pro department have been abolished, and about 100 other positions in the Siri and software teams have been cut. The move is part of the company's efforts to focus resources on new devices and artificial intelligence. People familiar with the matter said that in this adjustment, Apple has basically shut down a team dedicated to the Vision Pro game business, while also reducing the size of the department responsible for producing immersive video content for the device. Apple admitted in a statement that the company is making adjustments to some teams “to drive business development and provide the best experience for users.” [Castle Securities: Over 80% of the overall risk in the Situational Awareness Fund portfolio has been divested] According to the Financial Times, Castle Securities founder Ken Griffin responded to the company's acquisition of Situational Awareness assets under Leopold (Leopold) in a letter to clients on Friday. According to a letter obtained by CNBC, Griffin told clients that Castle Securities had divested more than 80% of the overall risk in the original purchased portfolio by conducting more than 100 major transactions (with a market value of more than $4 billion). In his letter, Griffin wrote, “A transaction of this scale would not have been possible without the full cooperation of the transaction teams and lead brokerage teams of the banks serving the two companies. I am very grateful for their dedicated efforts to complete the portfolio transfer quickly.” Griffin also confirmed that the company's flagship multi-strategy fund, the Wellington Fund, had a return of 5.94% in July, which is the fund's best monthly performance since 2022. [AI cloud company Nscale seeks to raise 3 billion US dollars in US IPOs] In comparison, AI cloud company Nscale is reportedly seeking to raise 3 billion US dollars in a US IPO. In the crypto market [Strategy stock price hit a two-month high, STRC returned above $96], the Bitcoin treasury company Strategy (MSTR) stock price rose to a two-month high today as the Bitcoin price briefly broke through $79,400. It broke through $120 during the intraday period, then partially regained its gains. Meanwhile, the price of STRC, Strategy's preferred stock product, also surpassed $96 for the first time since June. Previously, STRC's price once fell below $70 due to concerns about its ability to pay dividends and the ability of the stock price to maintain the $100 target for a long time. [Bernstein: Even if the Clarity Act is not passed, the SEC and CFTC will speed up rulemaking] Comparing news, the Bernstein analyst team led by Gautam Chhugani released a report stating that regardless of the procedural voting results of the “Clarity Act” on September 15, the certainty of US crypto regulation is expected to increase. They expect the SEC and CFTC to accelerate rulemaking in areas such as native crypto asset issuance, tokenized stocks, perpetual futures, computing power derivatives, and predictive markets. This regulatory clarity of expectations has become one of the broader supporting factors in the crypto market. 【A...

13h agoBitpushNews#Compare Daily Picks

The wave of AI infrastructure financing is competing with US bonds for long-term capital, and market concerns are driving up interest rate pressure

Comparing news, AI infrastructure investment is becoming a new variable in the US bond market. As tech giants expand the construction of data centers, chips, and computing power, AI companies' demand for financing grew rapidly, and they began to compete with the US government for capital from core bond buyers such as insurance companies, pensions, and long-term asset management institutions. According to the data, as of August, the issuance of US investment-grade corporate bonds reached about 1.7 trillion US dollars, a record high for the same period. According to Goldman Sachs data, the four major US technology companies have issued more than 170 billion US dollars in bonds since this year, which is more than the full year of 2025. Meanwhile, Broadcom is seeking chip and infrastructure financing for AI companies such as Anthropic, and the potential debt may be close to $100 billion. Market institutions pointed out that AI brought not only an increase in the supply of US bonds, but also a long-term expansion of supply in the entire bond market. When the government and technology companies simultaneously increase long-term financing needs, and the long-term capital pool is limited, the market may require higher returns to attract buyers. St. Louis Federal Reserve Chairman Mussalem said earlier that capital competition is forming between the US government's financing needs and AI infrastructure construction. Recently, the US bond market continued to be under pressure. The yield on US 30-year Treasury bonds once rose to 5.34%, a record high since 2007, and the 10-year US bond yield rose to 4.7%. The high interest rate environment is likely to further raise corporate financing costs and influence market pricing for AI companies through valuation discount rates. Meanwhile, US consumption data showed signs of weakness. Walmart's stock price fell about 9% in a single day, the biggest drop since 2022. The reason was that its same-store sales growth rate fell to its lowest level in six years, falling short of market expectations, indicating that consumer spending is slowing down. Against the backdrop of slowing economic growth and ongoing inflationary pressure, the Federal Reserve's policy faces a dilemma. The US Treasury recently expanded the scale of long-term US bond repurchases, raising the maximum single repurchase limit for 10-20- and 20-year US bonds from $2 billion to at least $4 billion. The market believes that the move is more of a signal. It has relieved the pressure on yield in the short term, but it has not changed the long-term supply and demand conflict. Analysts believe that future market attention will focus on US fiscal financing needs, AI capital expenditure expansion, and long-term interest rate trends. If long-term US bond yields continue to rise, the market may rediscuss policy tools such as yield curve control (YCC) or quantitative easing (QE). This article is sponsored by GENG, Build Your Fortune on GENG (https://geng.one)

1d agoburnking

DeepSeek visual model officially launched API: the price is exactly the same as V4 Flash

In comparison, DeepSeek officially launched a new visual model, deepseek-v4-flash-vision-exp. The official API documentation has juxtaposed it with V4 Flash and V4 Pro, and developers can directly import images through the DeepSeek API. The model supports 1 million token contexts, a maximum output of 384,000 tokens, and also supports JSON Output, Tool Calls, Responses API, and Anthropic API. The price is directly aligned with the V4 Flash. The peak period for each million token input is 3 yuan and the idle period is 1.5 yuan; the cache hit is only 0.1 yuan and 0.05 yuan, respectively. Each million tokens are worth 9 yuan during peak output periods and 4.5 yuan during idle periods. Compared to the V4 Pro, the price of the same class is only one-third. There is no separate charge per image. DeepSeek will convert the image into a token based on the image size, and then bill it together with the text token. Peak periods are 9:00 to 12:00 and 14:00 to 18:00 Beijing time; prices are halved the rest of the day.

1d ago

Anthropic economists decipher the AI unemployment paradox: skill-biased technology

Comparing news, Peter McCrory, head of economics at Anthropic, recently explained the AI unemployment paradox. He pointed out that AI capabilities have improved significantly, and it is reasonable to question why it has not had a significant impact on the unemployment rate. According to the US Business Trends and Outlook Survey, about one-fifth of US companies use AI to some extent, and productivity growth in the US has increased in recent years, some of which may be related to AI. McCrory said that the labor market unemployment rate is still close to the level of full employment determined by the Federal Reserve, and it is difficult to draw clear conclusions about the impact of AI even with in-depth data. The basic judgment is that up to now, AI has shown the characteristics of skills-oriented technology. It automates some work processes while complementing and amplifying professional abilities that rely on human participation, thereby creating maximum value in human-robot collaboration.

1d ago

Anthropic plans to adjust advanced model data retention policies to allow enterprises to self-host

According to Bloomberg, Anthropic plans to adjust the data retention policies of its most advanced artificial intelligence model to give enterprise customers greater control over their data. The company expects to launch a new security system later this year, and enterprise customers will still need to keep the data for 30 days, but they can choose to do it on their own cloud computing infrastructure instead of storing it on Anthropic servers. According to people familiar with the matter, the system has been in preparation for several months and has been developed in coordination with more than 100 customers in highly regulated industries. Anthropic announced in June that it will retain all customer data for 30 days for its more powerful Mythos, Fable models, and future cutting-edge products to identify and prevent new cyber attacks using its technology, while stating that the data will not be used for training. The company acknowledged in a recent report that the policy may not be popular with customers who are accustomed to zero reservations and may affect business. Anthropic declined to comment. Competitor OpenAI also recently began testing a solution called “private secure processing” to avoid customer data retention, and plans to launch it more widely in September.

1d ago