24-Hour AI News: AI Standards Implemented in the Financial Industry; Defense and AI Network Platforms Become the Focus

In the past 24 hours (October 30), China released a number of authoritative information regarding AI governance, industry scale, and filing management in the financial sector; the Asia-Pacific and North American sides saw new developments such as "defense × generative AI", "AI network platform", and open-source security models, presenting an overall pattern of policy implementation and infrastructure upgrades going hand in hand.

I. The People's Bank of China: Will expedite the issuance of policy documents related to "artificial intelligence + finance"

1. On October 30, the Science and Technology Department of the People's Bank of China clarified that it will implement the "Artificial Intelligence+" action plan and promote the standardized application of AI in the financial field.
2. Emphasize the "Evaluation Standard for Financial Application of Artificial Intelligence Algorithms" as a security framework and compliance baseline.
3. Focusing on three key implementation measures: risk control, algorithm transparency, and data governance.

2. "Financial Street Release" launches the "Application Guide to the Large-Scale Model of the Financial Industry"

1. On October 30, an authoritative institution released guidelines clarifying the boundaries and evaluation paths for banks, securities firms, and insurance companies using large-scale models.
2. Provide one-stop guidance from scenario selection and effect evaluation to compliance management.
3. Provide standardized references for financial digital transformation that is both "informed and compliant".

III. Hubei: The scale of the AI industry is expected to exceed 150 billion yuan for the whole year.

1. The press conference on October 29 revealed that the industry scale exceeded 110 billion yuan in the first three quarters, with an annual target of 150 billion yuan.
2. There are 1,215 AI-related enterprises and 12 provincial-level manufacturing innovation centers in the province.
3. Build a development system based on "technological innovation + industrial capacity + digital foundation + scenario application + ecosystem optimization".

IV. Beijing: A total of 162 generative AI services have been registered.

1. As of October 30, one new app was added that day, bringing the total number of apps registered to 162.
2. Applications that have already been launched must prominently display the names and registration numbers of the registered services and models they use.
3. The generated composite content must be labeled according to the specifications.

V. 360 Group: AI projects won in October totaled over 300 million yuan.

1. It was disclosed on the evening of October 29 that the Wuhan AI Innovation Application Demonstration Base Project (Phase I) won a bid of 132 million yuan.
2. Following the Hohhot project, another order worth hundreds of millions of yuan was secured.
3. The "Security × AI × Urban Scene" initiative is being accelerated.

VI. Lockheed Martin x Google Public Sector: Generative AI Enters the Localized Environment of National Defense

1. On October 29, a collaboration was announced, with generative AI such as Gemini being integrated into Lockheed's AI Factory.
2. The first phase will be implemented in a non-classified local environment, and will gradually expand to the "air gap" network.
3. The goal is to improve the efficiency of task planning, document processing, and knowledge retrieval.

VII. Cisco and NVIDIA Launch "AI-Ready" Networking Platform

1. Released on October 30, it launched the N9100 data center switch based on NVIDIA Spectrum-X and a unified AI architecture.
2. Provides reference architecture and AI-native wireless capabilities, covering multiple scenarios from cloud to enterprise to telecommunications.
3. To meet the high-bandwidth, low-latency interconnection requirements of large model training/inference clusters.

8. IBM Releases "Defense Model" for Defense Mission Planning

1. On October 29, the official announcement of the GA AI model for defense scenarios was made, focusing on mission planning and decision support.
2. Emphasizing controllability and compliance, serving national security and defense agencies.
3. Integrate with industry-customized data and processes to enhance the ability to extract reliable intelligence.

9. OpenAI launches open-source weighted security model family "gpt-oss-safeguard" (research preview)

1. Released on October 29th, an open-source weighted inference model for security classification and assessment.
2. Simultaneously launch the Open Community Collaboration (RMC) to encourage the co-construction of security toolchains.
3. Provide enterprises and research institutions with verifiable, secure and controllable compliance components.

Frequently Asked Questions (Q&A)

Q: Why is the financial sector releasing AI-related regulations and guidelines in such rapid succession?

A: First, the "AI + Finance" policy is about to be refined and implemented; second, industry guidelines provide the application boundaries and evaluation methods for large models; and third, risk control, transparency, and data compliance are taken into account to reduce the uncertainty of institutions introducing AI.

Q: What does the fact that Beijing has registered 162 generative AI applications mean?

A: This indicates that the accelerated standardization of model/service supply has entered the "licensed operation" stage; platforms and applications must publicize their filing information and label the generated content to facilitate social supervision and user identification.

Q: What are the key highlights of the Lockheed x Google collaboration?

A: Introduce generative AI such as Gemini into localized or even "air gap" environments to serve defense-related workloads in restricted networks; with a focus on improving the efficiency of document generation, question answering, and knowledge management within the security domain.

Q: What is the practical value of the Cisco x NVIDIA “AI-Ready” platform for enterprises?

A: We provide a replicable reference architecture and high-performance Ethernet interconnect to lower the barrier to building AI clusters; new products such as the N9100 are designed to meet the data traffic and jitter control needs of large model training/inference.