24-Hour AI News: SoftBank Makes a Big Bet, Australia Discusses AI Training with Copyright Framework, Sora Sparks Trust Controversy

In the past 24 hours (October 26), China has released new trends in the areas of "AI breeding", "open source of embodied world models", and "AI security and digital economy pilot zones"; North America and Australia have respectively seen large-scale capital increases for OpenAI and discussions on copyright frameworks, and disputes over trust and compliance in video generation on the application side have intensified.

1. SoftBank approves remaining $22.5 billion investment in OpenAI

1. On October 25, SoftBank's board of directors approved an additional US$22.5 billion to complete the previous financing arrangement totaling approximately US$41 billion.
2. This batch of funds is linked to OpenAI's governance and structural adjustments, with the goal of paving the way for subsequent capital market access.
3. Significance: Provides a long-term funding anchor for computing power and model iteration, strengthening the "capital-computing power-model" closed loop.

2. Australia launches discussions on AI copyright framework

1. On October 26, Australia announced that it would hold a two-day meeting starting from October 27 to evaluate the copyright adjustment path in the AI era.
2. Key signal: Exclude the "general exemption for text and data mining" and study "paid collective licensing" and solutions to reduce rights protection costs.
3. Impact: It may provide an operational paradigm for the “compliant commercialization of training data” and influence model training and content platforms.

3. Sora's rapid increase in short video app downloads sparks discussion of a "trust crisis" (North America)

1. On October 26, media reported that the app had been downloaded over one million times in its first week, sparking controversy over the creation of unauthorized characters and copyrighted materials.
2. Risk points: deep fakes, portrait and voice consent, and overflow of watermark removal tools.
3. Significance: "Generative AI - Content Security - Platform Governance" has become a common issue for regulators and the industry.

4. Joint study by 22 public media outlets: Chatbots have a high error rate in news responses (Europe)

1. A joint study released on October 26th stated that about 45% of the news questions and answers of four commonly used assistants had major problems.
2. The main flaws are in the citation of sources and the accuracy of facts, and the agency calls for strengthened independent monitoring.
3. Significance: News and public information scenarios place higher demands on “verifiable citations and fact-checking.”

5. "Fengdeng Gene Scientists" Accelerates AI Breeding (China)

1. On October 26, authoritative media reported that "Fengdeng Gene Scientists" carried out gene function prediction and experimental design assistance for multiple crops such as rice and corn.
2. The team said that their goal is to evolve into an intelligent breeding platform that "covers all species and the entire process."
3. Significance: AI+Agriculture has been implemented from models to "scientific research intelligence entities", shortening the cycle from hypothesis to verification.

6. WoW's embodied world model is open source, enabling robots to "preview the future" (China)

1. From October 25 to October 26, multiple media outlets reported that the Beijing Humanoid Robot Innovation Center and others released the open source world model WoW.
2. Features: Integrating vision, motion, and physical reasoning, it can make real-world operation predictions in scenarios such as homes, supermarkets, and industries.
3. Significance: Provides an open foundation for the “robot × world model” and promotes the expansion of embodied intelligence applications.

7. TechWorld Smart Security Conference Focuses on "Using Models to Treat Models" (Beijing, China)

1. On October 26, the industry conference focused on AI security, data security, and attack and defense confrontation.
2. Key points: Prompt word injection, data leakage, model loading, etc. have become core risks in large-scale model applications.
3. Implementation path: Build four types of security intelligence entities: content/data/computing power/business, and promote a standardized assessment and protection system.

8. Jiangsu Province was approved as a national digital economy innovation and development pilot zone and strengthened its "artificial intelligence plus" strategy (China)

1. On October 26, Jiangsu Province disclosed that the National Bureau of Statistics had replied agreeing to establish pilot zones in seven provinces and cities including Jiangsu.
2. A list of 158 reform items was released, of which Jiangsu was responsible for 31 items; emphasis was placed on the implementation of "artificial intelligence +" actions and high-value data scenarios.
3. Significance: Provides a policy testing ground for the industrial application of large models, the marketization of data elements and the coordination of regional computing power.

Frequently Asked Questions (Q&A)

Q: What is the biggest AI event on the global capital side today?

A: SoftBank has approved an additional $22.5 billion investment in OpenAI, which, combined with previous arrangements, constitutes a financing package of approximately $41 billion, binding the timing of governance and structural adjustments.

Q: What are the key points of Australia’s AI copyright discussion?

A: Instead of the "general TDM exemption", we will study the combination of "paid collective licensing + rights protection to reduce costs" and quickly clarify the policy options in a two-day meeting.

Q: What new developments in China’s scientific research are worth paying attention to?

A: "Fengdeng Gene Scientist" makes progress in crop gene function prediction and experimental design assistance; the open source of the WoW world model provides embodied intelligence with real-physics-oriented deduction capabilities.

Q: What are the main risks caused by Sora?

A: Consent for portraits and voices, boundaries of copyright re-creation, confusion of sources caused by watermark removal, and the resulting pressure on platform governance and public opinion trust.