news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

21st Century Business Herald reporter Xiao Xiao reports from Beijing

This week (August 12-17), there have been notable compliance incidents in overseas antitrust, domestic Internet content governance, and artificial intelligence.

In terms of anti-monopoly: the State Administration for Market Regulation is accelerating the formulation of the national standard "Regulations on the Management of Fair Competition and Compliance of Operators";

Personal information security: Home appliance company Ecovacs responded to questions about privacy leaks and said that smart home appliances lacked detailed regulations;

In terms of Internet content governance: During the rectification of "sports fan clubs", 5 administrative penalties have been imposed;

In terms of artificial intelligence: After receiving a "lawyer's letter warning" from CNKI, Mita AI will no longer include CNKI; restrict the surge in crawler websites;

Overseas: A startup uses algorithms to solve AI plagiarism problems; Microsoft's built-in Copilot was exposed to have data leaks; Google was exposed to consider splitting its business after losing the antitrust lawsuit; Universal Music and Meta expanded cooperation to negotiate AI music copyright issues

1. Antitrust

1. The State Administration for Market Regulation is accelerating the formulation of the national standard for the “Regulations on the Management of Fair Competition and Compliance of Operators”

On August 16, the State Council Information Office held a series of press conferences on "Promoting High-quality Development". At the conference, the State Administration for Market Regulation revealed that this year, it will further promote the normalization of anti-monopoly supervision in the digital economy and other fields, and unconditionally approve 48 cases of concentration declaration by platform business operators in accordance with the law. It is currently accelerating the formulation of the "Regulations on Fair Competition and Compliance Management of Operators".

2. Personal Information Security

1. Ecovacs responds to privacy leak questions, smart home appliances lack regulatory details

Recently, two security researchers said at the Def Con security conference that they found security issues in Ecovacs' sweeping robot products. After connecting to the Ecovacs robot via Bluetooth, hackers can remotely control it through the product's built-in WiFi connection function and access functions and information such as room maps, cameras, and microphones in its operating system.

In response to the above questions, Ecovacs said in an interview with Southern Finance Omnimedia that data security and user privacy are among the issues that Ecovacs attaches the most importance to. The Ecovacs Robotics Safety Committee has conducted internal research and reviews on issues such as network connection and data storage, and concluded that these security risks are extremely unlikely to occur in the daily use environment of users, and require professional hacking tools and close contact with the machine to complete, so users do not need to worry about it. Despite this, Ecovacs will actively optimize its products based on research and review findings.

Southern Finance Comment: Behind the frequent safety issues of various smart home appliances, on the one hand, the current situation of corporate security construction needs to be further improved, and on the other hand, there is a lack of regulatory details in related fields.Although my country currently has regulatory requirements in areas such as network security and hardware design and manufacturing, it has been lacking corresponding subdivision standards in the field of intelligent products that combine software and hardware, and there is no way to talk about various security requirements and safeguards extended on this basis.

III. Internet Content Governance

1. During the rectification of "sports fan clubs", 5 administrative penalties have been imposed

After the Paris Olympics women's singles table tennis final, four platforms, Weibo, Douyin, Toutiao, and Kuaishou, released reports on the governance of "uncivilized viewing" during the Olympics. Douyin and Kuaishou used the term "fan circle" in their titles to describe these chaos. Each platform identified no less than 20,000 pieces of inappropriate speech, and hundreds of accounts were banned. Weibo opened a "fan circle complaint area" specifically for this purpose last week.

Public security organs are also intervening in the governance. On August 6, the public security organs in Daxing, Beijing, found that a fan had spread rumors about the relationship between athletes and coaches, and the fan had been administratively detained; on August 15, the Ministry of Public Security once again announced four typical cases of cracking down on and rectifying illegal and criminal activities in "fan circles" in the sports field. The public security organs in Guangdong, Shandong, Hebei, and Henan have all taken compulsory administrative penalties.

Nancai Comment: Fan circles are organizations with clear different voices and hierarchical levels of power, which can have a huge impact. The scholars interviewed believe that fan circles cannot be simplified as a gathering of fans, but they are not inherently a problem. Fan circles are the product of the demands of the sports industry, platform technology, and entertainment capital. Only by recognizing this can we talk about governance issues.

4. Artificial Intelligence

1. The surge in websites that restrict crawlers has intensified the contradiction between data supply and demand

Recently, artificial intelligence expert Andrew Ng mentioned a study on data licensing on the website The Batch. The study found that various websites crawled by open source data sets such as C4, RefineWeb, and Dolma are rapidly tightening their licensing agreements. Since the emergence of GPTBot (mid-2023), the number of websites that are completely restricted at the robots.txt level has surged. OpenAI, Anthropic, and Common Crawl are ranked in the top three in terms of restrictions, all reaching more than 80%, but website owners are generally more tolerant and open to crawlers in non-AI fields such as Internet Archive or Google Search.

Researchers are concerned that this will not only affect the training of commercial AI models, but will also hinder research in academia and non-profit organizations.

2. After receiving a "lawyer's letter warning" from CNKI, Mita AI will no longer be included in CNKI

Recently, CNKI accused "Mita AI Search and Mita AI Search APP of providing users with our company's academic literature titles and abstract data in huge quantities. This behavior was done without our company's permission and seriously infringed our company's legitimate rights and interests." CNKI therefore requested that China National Knowledge Infrastructure do not want to be searched by Mita Technology and required Mita AI to immediately disconnect the search results.

On August 16, Mita AI responded to HowNet, emphasizing that it only included abstracts and titles of papers, not the content of the articles themselves, and that reading papers still requires jumping to the website to obtain them. Mita AI said it did not understand but respected HowNet's choice, and would no longer include titles and abstracts of HowNet documents from today, and would instead include data from other authoritative Chinese and English knowledge bases.

Nancai Comment: CNKI's accusation is mainly aimed at two contents: document titles and abstracts. The simple title and directory are generally not considered to constitute works (in the sense of copyright) and do not involve infringement issues. However, the abstract part is very likely to be identified as a work, and unauthorized citation may constitute infringement.

It is worth mentioning that at a forum attended by 21 reporters last month, a senior executive of CNKI revealed that CNKI has recently begun to explore ways to monetize data and has reached a cooperation with Huawei.

5. Overseas

1. A startup uses algorithms to solve AI plagiarism problems

On August 6, the startup ProRata.AI was established in California. Its founder is Bill Gross, who once proposed paid ranking advertising and advertising pay-per-use. Now Bill Gross has proposed a business model for the AI ​​era: AI pay-per-use. Relying on a patented algorithm of the company, the output of AI will be disassembled into different parts, the corresponding copyright source will be found, and the income will be distributed according to the output ratio.

Although there is no fully implemented product yet, major content copyright holders such as the Financial Times, Fortune, and Universal Music Group have signed cooperation agreements with ProRata. The company has currently received $25 million in Series A funding.

2. Microsoft's built-in Copilot was exposed to data leakage

A researcher at the Black Hat 2024 conference revealed a surprising discovery that Microsoft's AI assistant Copilot has multiple security vulnerabilities that attackers can exploit to steal sensitive data and even turn it into a powerful phishing attack tool. The findings show that targeting thousands of accessible AI assistants can reveal sensitive data and company credentials that malicious actors can exploit.

In addition, the researchers demonstrated how attackers can use these vulnerabilities to launch attacks, tricking Copilot into modifying bank transfer information without having to gain access to corporate accounts, and even without having to get the target employees to open the email to launch the attack.

Nancai Comment: Copilot has two main problems: First, the training data will inevitably contain private information. When AI is given access to data, the data becomes an attack surface for prompt injection. To some extent, if a robot is useful, it is vulnerable; if it is not vulnerable, it is useless. This is a fundamental contradiction; in addition, the interaction with the public cloud also increases the risk.

In response to the security risks brought about by the development of AI technology, Zhang Yaqin, academician of the Chinese Academy of Engineering and dean of the Institute of Intelligent Industries at Tsinghua University, suggested that companies or institutions engaged in cutting-edge large models should invest 10-30% of their investment in related research or product development.

3. After losing the antitrust lawsuit, Google was reported to be considering splitting its business

On August 13th local time, foreign media quoted people familiar with the matter as saying that after a US court ruled last week that Google's search business violated US antitrust laws, the victorious US Department of Justice has considered the rare punitive measure of breaking up Google.

According to people familiar with the matter, if the Justice Department pushes forward with the split plan, the most likely units to be spun off from Google are the Android operating system and the web browser Chrome. Once the Justice Department's Google split plan is implemented, it will become the largest U.S. corporate split since the U.S. telecommunications company AT&T was split in 1984.

4. Universal Music and Meta expand cooperation to negotiate AI music copyright issues

On August 12, Universal Music Group announced an expansion of its multi-year partnership agreement with Meta, which allows users to share songs from the Universal Music library on Meta's platforms (Facebook, Instagram, Horizon, Threads and WhatsApp) without infringing copyright. The most striking thing about the new agreement is that it states that the two companies are addressing the issue of "unauthorized AI-generated content."

Earlier this year, Universal Music broke up with TikTok. In a statement released in February, Universal Music emphasized that the company was concerned that the platform's AI technology threatened the artists' copyright pool. The two companies reached a truce agreement in May, and the music copyrights of artists such as Taylor Swift returned to TikTok.