AI Weekly | ByteDance big model training is "poisoned"; Microsoft will terminate China's personal Azure OpenAI service
王俊杰2017
发表于 2024-10-20 11:05:31
168
0
0
ByteDance big model training is "poisoned"
Recently, it was reported that ByteDance was "poisoned" by interns in large model training. It is reported that the event occurred in the ByteDance commercial team. Because the intern Tian was dissatisfied with the team resource allocation, he used the HF (hugging face) vulnerability to inject broken code through the shared model, resulting in damage to the team model training results. On October 19, ByteDance released a clarification statement that the involved interns maliciously interfered with the model training task of the research project of the commercialization technology team, but it did not affect the formal commercial project and online business, nor did it involve other businesses such as the ByteDance big model. The online rumor that it involves over 8000 cards and losses of tens of millions of dollars is severely exaggerated. After verification, the person involved has been interning in a commercial technology team and has no experience in AI Lab internships. ByteDance said that the intern had been dismissed by the company in August. The company will also synchronize its actions with the industry alliance and the school where it is located, and hand them over to the school for handling.
Comment: This event exposed the security management problems existing in the technical training of ByteDance, including permission isolation, audit of shared code, etc. An industry insider stated that implementing permission isolation and auditing is beneficial for protecting the company's core data and intellectual property, preventing data leakage, and improving the security of data and systems. For example, real-time monitoring of permission usage, timely detection of permission abuse and abnormal operations; Regularly audit permissions, check whether team members' permissions comply with authorization policies, and whether there is any abuse of permissions. But this matter also has difficulties, including cross departmental cooperation costs, regular maintenance and update resource investment, etc.
Microsoft will terminate China's personal Azure OpenAI service
On October 17, several developers received an email from Microsoft saying that due to local regulatory requirements, Microsoft Azure OpenAI service will be closed on October 21, and only enterprise customers can subscribe to Azure OpenAI service in Chinese Mainland in the future. Some developers have stated that "this is the only window for individuals to use OpenAI in compliance in China, and it has now been closed." At the same time, some developers are dissatisfied with Microsoft's "giving four days for processing. Regarding this matter, Microsoft has not yet made a statement.
Comment: On June 25th, OpenAI sent an email to some domestic developers stating that it will terminate its API services to China. At that time, Microsoft Technology had also written articles to teach users how to migrate to Microsoft Azure OpenAI. According to the agreement signed between OpenAI and Microsoft, OpenAI can receive a 20% share of the revenue from Microsoft's sale of OpenAIGPT services. At that time, there were voices suggesting that Microsoft would be the biggest winner after OpenAI terminated its API services to China. Just four months later, things changed. Regarding the termination of Azure services this time, some developers have expressed that it is forcing individuals and small and medium-sized developers to use the large models in China.
Google AI Team Architecture Adjusted Again
After Google merged its two major AI labs, Google Brain and DeepMind, in April this year, the architecture of Google's AI team is undergoing further adjustments. On October 17th local time, Google officially released an open letter written by Google CEO Sundar Pichai, which mentioned merging the Gemini App application team into Google DeepMind to further integrate the AI team. The assistant team focused on home products will merge with the platform and device department. At the same time, Prabhakar Raghavan, Senior Vice President responsible for important businesses such as search and advertising, will be appointed as Chief Technology Officer, while Nick Fox will take over the position of Head of Search and Advertising.
Comment: This adjustment is considered as another move by Google in the field of artificial intelligence. At the same time, Google's search business is facing numerous crises. Previously, it was ruled by the court as a market monopoly and faced the risk of being split. Fox seems to have high expectations. The development speed of artificial intelligence is faster than any previous technology. In order to continue accelerating the pace of progress, we have been working hard to simplify our structure, including creating Google DeepMind and joining the platform and device teams, "Pichai said.
Amazon invests $500 million to develop nuclear energy
On October 16th local time, Amazon announced that it will invest over $500 million to develop small modular nuclear reactors (SMRs) to meet the huge demand for clean energy when expanding its cloud computing services to generative AI, and as part of achieving net zero carbon emissions. Amazon Web Services (AWS) has reached an agreement with Virginia utility Dominion Energy to explore the development of SMR near the existing North Anna nuclear power plant. Amazon has also reached an agreement with Energy Northwest to fund the development of four SMRs in Washington state, which will directly supply power to the grid and help meet the energy needs of Amazon's operations.
Comment: Among major companies, besides Amazon, Google and Microsoft have also shown strong interest in nuclear energy. Google has obtained a new nuclear energy contract, while Microsoft plans to restart the Three Mile Island nuclear power plant before 2028. This trend demonstrates the energy anxiety of technology companies and their efforts to pursue sustainable development.
Nvidia's stock price once hit a historic high
Boosted by TSMC's third quarter financial report, several semiconductor stocks in the US stock market rose on Thursday. Nvidia's stock price hit a historic high of $140.89 per share during trading, but the closing price increase narrowed to $136.93 per share. Since September this year, Nvidia's stock price has started to fluctuate and rise. Billionaire investor and founder of the Duken family office, Stanley Druckenmiller, stated this week that his decision to sell off Nvidia stock earlier this year was a "major mistake".
Comment: The outside world is quite concerned about the cooperation and binding relationship between AI chip giant NVIDIA and contract manufacturer TSMC. The smooth production and delivery of Nvidia Blackwell architecture chips may simultaneously affect the future performance of both companies. Previously, it was reported that the production of Nvidia Blackwell chips has been delayed recently. To meet the needs of customers, TSMC is maintaining a high level of capital expenditure.
Social platform X allows third parties to use user data to train AI
On October 17th local time, social networking platform X updated its privacy policy, allowing third-party "partners" to use X data to train AI models if users choose not to exit. This change means that X may, like Reddit and multiple media organizations, open up new revenue channels by authorizing data to AI companies. The new policy will take effect on November 15th, at which time users can choose to opt out of data sharing.
Comment: Currently, the new privacy policy has not clearly stated where users should set up data sharing. The new privacy policy will take effect on November 15th, and relevant options may be set at that time. Previously, there were reports that the tweet content of X's users was used to train the Grok AI model. Allowing third parties to use user data to train AI may bring new revenue streams for X.
Chip companies say 70% increase in semiconductor market is related to AI
At the SEMiBAY exhibition held from October 16th to 18th, several industry insiders proposed that the global semiconductor market size is expected to exceed one trillion US dollars by 2030. ASML Marketing Director Tao Tingting stated that the driving force for the growth of the semiconductor market in the next decade will come from the AI, automotive, and industrial sectors. Dai Weimin, founder, chairman, and president of Xinyuan Microelectronics, believes that in the process of the semiconductor industry's scale moving towards trillions of dollars, 70% of the increment will be related to AI, and semiconductor companies unrelated to AI can only receive the remaining 30%.
Comment: This round of semiconductor market recovery is not only driven by cyclical factors, but also by industry changes. In the past decade, a major driving force for the growth of the semiconductor industry has come from mobile phones, PCs, and servers, and now it comes from fields such as AI and automotive. A major driving force for the future replacement of the mobile phone and PC markets also comes from AI.
Ministry of Industry and Information Technology: The registered users of China's generative artificial intelligence service model have exceeded 600 million
Recently, the Ministry of Industry and Information Technology stated that the number of registered users of China's generative artificial intelligence service model has exceeded 600 million. According to Zhao Zhiguo, chief engineer of the Ministry of Industry and Information Technology, the scale of China's AI core industry is constantly improving, with more than 4500 enterprises. Nearly 200 generative artificial intelligence service models have been registered and launched to provide services to the public, with over 600 million registered users.
Comment: This year is a year of accelerated implementation of big models, especially for platform based enterprises that promote the application of big models, which may have driven a lot of user volume. Zhao Zhiguo mentioned that in the future, we need to enhance the ability of artificial intelligence infrastructure to provide public services, build a low threshold, easy to adapt, and highly available artificial intelligence development platform, and accelerate the training, optimization, and large-scale application of large models. At the same time, we will enhance the supply capacity of computing power facilities, accelerate the construction of a nationwide integrated computing power system, and steadily increase the proportion of intelligent computing power.
After TSMC releases its third quarter financial report, its market value ranks among the top eight in the world
Boosted by TSMC's third quarter financial report, the US stock market closed up 9.79% on Thursday, with a market value of $1.07 trillion at the close. According to the market value ranking website Companies MarketCap, TSMC's market value has surpassed Warren Buffett's Berkshire Hathaway, becoming the eighth largest company in the world by market value, second only to Apple, Nvidia, Microsoft, Google, Amazon, Saudi Aramco, and Meta. TSMC's consolidated revenue for the third quarter was $23.504 billion, a year-on-year increase of 39.0%, and its net profit was $10.058 billion, a year-on-year increase of 54.2%.
Comment: TSMC released a signal at its earnings briefing after the release of its financial report that the demand for AI remains strong. The management of TSMC stated at the performance exchange meeting that the demand for AI is real, and the demand growth is just beginning, and the current demand is crazy. The proportion of revenue from TSMC's advanced processes has also increased, with 7nm and below advanced processes accounting for 69% of Crystal Circle's total revenue in the third quarter, higher than the 67% and 65% in the second and first quarters, respectively. At present, there is a rigid demand for advanced processes in fields such as mobile phones and AI.
CandyLake.com 系信息发布平台,仅提供信息存储空间服务。
声明:该文观点仅代表作者本人,本文不代表CandyLake.com立场,且不构成建议,请谨慎对待。
声明:该文观点仅代表作者本人,本文不代表CandyLake.com立场,且不构成建议,请谨慎对待。
猜你喜欢
- The Nasdaq China Golden Dragon Index fell more than 1%, while Xiaopeng Motors fell more than 3%
- Starbucks China responds after 24 years of store lease ends! Fourth quarter revenue decreased by 7% year-on-year
- Most popular Chinese concept stocks fell on Monday, and the Nasdaq China Golden Dragon Index fell more than 2%
- Wal Mart stores in China are fully launched in Meituan
- Most popular Chinese concept stocks in the US stock market have risen, while the Nasdaq China Golden Dragon Index has risen by nearly 2%
- Eli Lilly's Alzheimer's disease therapy, Minolta (Donaimab Injection), has been approved in China
- Yum Brands China: The number of domestic stores with 30 Pizzahut products has reached 3606
- Eli Lilly's Alzheimer's disease therapy, Minolta (Donaimab Injection), has been approved in China
- BeiGene ADC's new cancer drug BG-C137 has been clinically approved for the first time in China
- Popular Chinese concept stocks fluctuated on Friday, with the Nasdaq China Golden Dragon Index closing up 0.53%
-
隔夜株式市場 世界の主要指数は金曜日に多くが下落し、最新のインフレデータが減速の兆しを示したおかげで、米株3大指数は大幅に回復し、いずれも1%超上昇した。 金曜日に発表されたデータによると、米国の11月のPC ...
- SNT
- 前天 12:48
- 支持
- 反对
- 回复
- 收藏
-
長年にわたって、昔の消金大手の捷信消金の再編がようやく地に着いた。 天津銀行の発表によると、同行は京東傘下の2社、対外貿易信託などと捷信消金再編に参加する。再編が完了すると、京東の持ち株比率は65%に達し ...
- SNT
- 前天 12:09
- 支持
- 反对
- 回复
- 收藏
-
【GPT-5屋台で大きな問題:数億ドルを燃やした後、OpenAIは牛が吹くのが早いことを発見した】OpenAIのGPT-5プロジェクト(Orion)はすでに18カ月を超える準備をしており、関係者によると、このプロジェクトは現在進 ...
- SNT
- 5 小时前
- 支持
- 反对
- 回复
- 收藏
-
【ビットコインが飛び込む!32万人超の爆倉】データによると、過去24時間で世界には32万7000人以上の爆倉があり、爆倉の総額は10億ドルを超えた。
- 断翅小蝶腥
- 3 天前
- 支持
- 反对
- 回复
- 收藏