AI Daily News — 2026-09-24 (Thu) Morning News

AI industry trend report (last 12 hours)

Overview

The CEOs of OpenAI and Anthropic gave an unprecedented briefing on AI governance at the UN Security Council, suddenly elevating the AI safety debate to a geopolitical level. On the other hand, Jensen Huang directly refutes the "AI apocalypse", and the division between "decelerationists" and "accelerationists" within the industry is becoming clearer. Google introduced the TTS model of Gemini 3.8, and voice AI has emerged as a new competitive axis.

Anthropic / Claude trends

Dario Amodei and Sam Altman briefed the UN Security Council on AI risks and called for international cooperation. Rest of World points out geopolitical tensions, saying, ``Amodei wants to slow down AI, but China won't follow his instructions.'' On the business side, Amazon is integrating Claude as an outside AI agent through Seller Assistant and new plugins, and implementation in the e-commerce space is progressing.

OpenAI trends

Sam Altman called for the creation of global AI standards at the United Nations Security Council, contrasting his stance with the Trump administration's rejection of the "globalist vision of AI domination." Mashable reports that the IPO has been postponed due to safety concerns. On the technical side, an OpenAI agent solved an extremely difficult mathematical problem, but the ``proof'' was in a unique format that was difficult for even mathematicians to understand, which became a hot topic. The company is also actively responding to competition, including developing new features to compete with Grok Bot and Meta's Muse, and filing a motion to dismiss Musk's AI antitrust lawsuit.

Google AI trends

Gemini 3.8 Flash TTS and Flash-Lite TTS have been announced, strengthening the offensive in the speech synthesis field. Experian becomes the first personal finance app to integrate Gemini, and also adds support for local AI models in its Antigravity SDK. On the security front, on the other hand, it was reported that Gemini AI can be used to hack companies by guessing passwords, highlighting the dual nature of AI.

[Trends of other AI companies] (Meta, Microsoft, xAI, DeepSeek)

Meta has seen its companion AI app Muse rise to the top of the charts, and is expanding its AI shopping capabilities through a partnership with Shopify. While Microsoft's Security Copilot is becoming more popular, the Copilot browser is facing some backlash from users. xAI continues to grow rapidly, with the number of weekly Grok users reaching 418,000, but a lawsuit has been filed alleging that CSAM (child sexual abuse material) was used in Grok's training, creating a major reputational risk. It was reported that DeepSeek is testing a more efficient and secure AI agent training method, but at the same time it was also revealed that Chinese hackers exploited Anthropic, DeepSeek, and Moonshot AI agents to carry out cyberattacks on 100 companies.

Remarks and trends of AI celebrities

- Sam Altman (OpenAI CEO): Advocates for the creation of global AI standards at the United Nations Security Council. There are also reports of IPO postponement. - Dario Amodei (Anthropic CEO): Briefing with Altman at the UN Security Council. Although China insists on slowing down AI development, it has been pointed out that China is not following suit. - Elon Musk (xAI Founder/Tesla CEO): Unconventional statement praising China's leadership and calling for stricter AI regulations before a state dinner. Another commenter agreed with Amodei's slowdown theory, saying, ``Dario is right.'' - Jensen Huang (NVIDIA CEO): ``All predictions have been wrong,'' dismissing the AI apocalypse. He has clearly stated his opposition to exempting AI companies from regulations, and has also said, ``Don't ship products (in case things go out of control).'' - Sundar Pichai (Google/Alphabet CEO): Analysis shows that Alphabet's Agentic AI "Spark" will outperform Meta's "Muse" in the long term. - Mark Zuckerberg (Meta CEO): Given the popularity of Muse, announced "further partnerships" with Shopify and others. - Yann LeCun (Meta Chief AI Scientist): Points out that humanoid robot development is stalled due to lack of learning data. - Geoffrey Hinton (AI Godfather): In media appearances, he presents his theories on how to create "AI that will not destroy humanity."

Breakthroughs and innovations in AI technology

An OpenAI agent solved one of the most difficult mathematical problems, but the proof process was considered difficult to decipher even by the mathematician community, drawing renewed attention to the black box nature of AI's reasoning process. DeepSeek is testing a new method for training AI agents that is both efficient and secure, and could be an answer to the common industry challenge of balancing cost efficiency and safety.

AI Regulation/Policy/Social Impact

The top topic was a UN Security Council briefing by the CEOs of OpenAI and Anthropic, which comes as the Trump administration rejects the idea of "globalist AI domination." In California, Governor Newsom announced a team of experts that will include implementing a "kill switch" as part of his AI executive order. Italy is moving forward with implementation of the EU AI Act, and the state of Maryland has announced an AI framework that includes restrictions on tax benefits for data centers, and the differences in regulations at the state and national level are widening.

Integrated analysis: What is happening now

The biggest structural change is that the AI safety debate has been elevated from a competitive issue to a national security issue. The simultaneous briefing by Altman and Amodei at the United Nations Security Council is symbolic, and the conflict between Amodei and Musk's desire to "decelerate" and Huang's desire to "continue accelerating" is manifesting as an ideological divide in the industry. At the same time, "AI safety" is coming to the fore not only as a matter of ideological debate, but also as a concrete risk that can cause real harm, such as DeepSeek's search for safety training methods, Chinese hackers' abuse of AI agents from multiple vendors, and xAI's CSAM lawsuit. On the business side, the battle for position in the agentic AI/AI commerce field is accelerating in the form of Amazon x Claude, Meta x Shopify (Muse), Google Spark, and OpenAI's ability to compete with Grok/Muse. Also, it cannot be overlooked that voice AI (Gemini TTS) has emerged as a new axis of differentiation. In terms of geopolitics, the contrast between Musk's praise for China during his visit to China and Amodei's warning against China has made the difference in enthusiasm between the US and China in AI development even clearer.

Future points of interest

Will an effective global AI governance framework be formed after the UN Security Council briefing, and how will the conflict with the Trump administration be resolved? 2. Battle for leadership in companion/agentic AI competition using competing functions of Meta “Muse”, Google “Spark”, and OpenAI. 3. Development of xAI CSAM litigation and increasing trend of safety-related litigation against AI companies.