24-hour AI news: The implementation of energy "AI+" has accelerated, and Anthropic has debuted on both fronts

24-hour AI news at a glance: domestic policies and industries "double force", overseas supervision and enterprise trends are upgraded simultaneously In

the past 24 hours, China has released "artificial intelligence +" implementation opinions in the energy field, local industries and rule of law supporting new moves; On the overseas side, cutting-edge companies have blossomed in regulatory compliance, technology releases and landing orders, and maintained balanced coverage of domestic and foreign quantities and importance.

1. The two departments issued "artificial intelligence + energy" implementation opinions

1. Issued on September 8, it was proposed to initially build an energy-intelligence integration innovation system by 2027, and promote the in-depth application of more than 5 professional large models in power grids, power generation, coal, oil and gas and other industries.
2. Clearly form more than 10 replicable key demonstration projects, explore 100 typical application scenarios and improve about 100 technical standards.
3. Significance: Consolidate the foundation of computing power-power synergy, clarify medium-term milestones, and drive the investment and standardization process of the energy AI industry chain.

2. Shanghai has introduced 12 measures to support the "AI + advertising industry"

1. On September 8, the "Several Measures" were issued, proposing to support the research and development of "AI + digital advertising" agents and the application of basic models.
2. The total support of eligible projects at the urban level shall not exceed 12 million yuan; The maximum support will not exceed 5 million yuan according to the same proportion of R&D investment.
3. Significance: Provide funding and scenario guarantee for the compliant application of generative AI in the content production and delivery link.

3. Chongqing Institute of Artificial Intelligence and General Artificial Intelligence Research Institute unveiled

1. On September 8, Chongqing announced the unveiling and signed a cooperation agreement to promote school-local collaborative education and scientific research transformation.
2. Positioning: focusing on key technologies, industrialization and talent training of general artificial intelligence (AGI).
3. Significance: The AI higher education and industry-research integration platform in the western region will be increased to serve the local "intelligent manufacturing" upgrade.

4. The China-ASEAN Expo focuses on AI and sets up the first "exhibition agent"

1. According to the information during the conference on September 8, nearly 200 AI companies and start-up teams made a concentrated appearance and held more than 10 AI-themed activities.
2. The first domestic "AI exhibition agent" was launched, nearly 70 blockbuster new products were released, more than 80 economic and trade activities were held, and 100+ major projects of digital economy and green economy were expected to be signed.
3. Significance: The regional economic and trade expo scene takes the lead in AI to promote cross-border AI industry cooperation and order conversion.

5. The Supreme People's Court issued typical cases against unfair competition, involving the protection of algorithms and model elements

1. Eight typical cases were released during the "Fair Competition Policy Publicity Week" on September 8.
2. According to media interpretation, cases related to the digital economy involve the expression of judicial rules for the competition order of new factors such as algorithms, model parameters and structures.
3. Significance: Provide clearer judicial guidance for the competitive boundaries and compliance applications of AI models.

6. Anthropic publicly supports California's SB 53 Frontier AI Regulation Act

1. On September 8, it expressed its support for SB 53, advocating a "trust but verification" disclosure and transparency mechanism.
2. Key points: For powerful models above the 10^26 FLOPs training threshold, security frameworks, transparent reporting, reporting of key events within 15 days, and whistleblower protection are required.
3. Significance: The head laboratory endorses the state-level "light compulsion and heavy disclosure" path, or affects the legislative orientation of other jurisdictions.

7. According to reports: Anthropic and author class action lawsuit reached a $1.5 billion settlement

1. On September 8, a number of media said that the two sides reached a settlement arrangement totaling about $1.5 billion (USD), which is still pending confirmation by the court process.
2. Focus: Training data copyright compliance and compensation mechanism, or become a sample for negotiations between the content industry and AI platforms.
3. Significance: High settlement promotes the three-piece set of "authorization-compensation-transparency" to become a normalized issue in the industry.

8. Fujitsu releases generative AI "refactoring" technology for lightweight and efficient models

1. Released on September 8, as a core component of Kozuchi AI services to enhance Takane LLM and achieve more power-saving lightweight model generation.
2. Goal: reduce power consumption and inference costs while maintaining performance, and facilitate industrial implementation.
3. Significance: The engineering route from large to "clever" provides reusable technical parts for "AI cost reduction and efficiency increase".

9. BigID launches MCP connectivity for natural language reports and compliance insights

1. Launched on September 8, it supports collaboration with Copilot, Gemini, Claude, etc., and users can generate data reports and audit points in natural language.
2. Positioning: AI-native analysis and report automation in the field of data governance and privacy compliance.
3. Significance: The combination of compliance Tech and generative AI has been further accelerated, and enterprise data security and audit processes have been AI-enabled.

10. Gorilla Technology won an order for an Asian government-level AI intelligence platform

1. On September 8, it announced that it had won the bid for an IPDR-based AI intelligence analysis platform in APAC, which is its second national-level AI law enforcement deployment project in 2025.
2. Significance: The large-scale application of AI in national governance and law enforcement scenarios such as network recording (IPDR) has accelerated, highlighting the traction of security compliance demand.

Frequently Asked Questions (Q&A)

Q: What are the quantitative goals for "AI+" in the energy sector by 2027?

A: The official goals include: promoting the in-depth application of more than 5 professional large models, forming more than 10 key demonstration projects, exploring 100 typical application scenarios, and improving about 100 technical standards (released as of September 8).

Q: What specific requirements does California SB 53 require for cutting-edge AI developers to meet their transparency and security obligations?

A: For large-scale model developers with training computing power of 10^26 FLOPs, they need to issue a security framework and transparency report, report key security incidents within 15 days, and set whistleblower protection and breach penalties (key points announced on September 8).

Q: Has Anthropic's $1.5 billion settlement been finalized?

A: At present, the media report has reached an arrangement of about US$1.5 billion (USD), which still needs to be confirmed by the court and disclosed in detail, which is in the category of "news claimed/to be confirmed".

Q: What pain points does Fujitsu's "generative AI reconstruction technology" solve?

A: It significantly reduces power consumption and model size while maintaining performance, making it easier for enterprises to land in edge and cost-sensitive scenarios.

Q: What is the most direct benefit of Shanghai's "AI + Advertising" new policy to start-ups?

A: The two-level financial support and the direction of "agent" are clear, with a maximum support of 12 million yuan for a single project, and a maximum of 5 million yuan for R&D investment to meet the standard, reducing the threshold for trial and error and scale.