On July 6, the United Nations launched its first session of the "Global AI Governance Dialogue" in Geneva, marking the establishment of the world's first international platform for AI governance. This initiative brings together all 193 UN member states along with representatives from the private sector, academia, and civil society.
In his opening remarks, UN Secretary-General António Guterres emphasized the need for a global governance framework to ensure that AI does not dictate humanity's future. He warned that high-performance AI chips designed for civilian use are increasingly being repurposed for military applications, leading to the normalization of "killer robots," and called for extensive international regulation.
The dialogue incorporates an evaluation report from an independent international scientific panel co-chaired by Yoshua Bengio and Maria Leza. The event is part of a series of gatherings, including the WSIS Forum (July 6-10) and the ITU "AI for Good" Global Summit (July 7-10), with a second session scheduled for May 2027 in New York.
Altman Proposes US-Led International AI Oversight Forum
Sam Altman, CEO of OpenAI, proposed the creation of a US-led international forum to establish AI safety standards and oversee research institutions in a Financial Times op-ed. He cited models such as aviation safety, international financial standards, and the International Atomic Energy Agency (IAEA). Concurrently, OpenAI is discussing a plan to offer the US government a 5% stake in the company, valued at approximately $42.6 billion. Major AI developers like Anthropic, Google, and Meta are also considering contributing 5% to a sovereign wealth fund modeled after Alaska's Permanent Fund.
According to the Financial Times, this move aims to secure a relationship with the administration and mitigate political backlash, although actual implementation may require congressional approval. The stake proposal emerged less than a week after OpenAI agreed to limit the release of GPT-5.6 to a select group of partners, raising concerns among governance scholars that holding stakes in regulated companies could weaken enforcement efforts.
Washington Post Reports on Secret AI War Between Silicon Valley and China
On July 6, the Washington Post published an in-depth report alleging that US tech companies are accusing Chinese competitors of attempting to "distill" their chatbots. US firms claim that China is effectively using their chatbots as "tutors" to train its own AI systems. This follows Anthropic's public condemnation of Alibaba's campaign to extract capabilities from its Claude model, indicating that the US-China AI rivalry is escalating over model access control and intellectual property issues.
Alibaba Restructures AI Division as Token Processing Surges
Alibaba has restructured its AI division, consolidating five organizations, including Tongyi Lab and Qwen, under the newly formed "Alibaba Token Hub," which reports directly to CEO Eddie Wu. China's daily token processing volume has surged from 100 billion tokens in early 2024 to 140 trillion currently, with Chinese AI accounting for approximately 45% of total traffic on OpenRouter. Meanwhile, China's "Interim Measures for the Management of AI Humanization Interaction Services" will take effect on July 15, leading Doubao, the largest AI app in China with 345 million monthly users, to suspend its agent features. The regulations mandate addiction prevention systems, usage time notifications, and immediate termination mechanisms for AI services mimicking human personalities, while banning virtual partner and family services for minors.
Anthropic Ends Free Access to Claude 5, Shifts to Usage-Based Billing
July 7 marks the last day for free access to Anthropic's top model, Claude 5, which has been included in Pro, Max, Team, and select enterprise plans. Starting July 8, usage-based billing will be implemented at a standard API rate of $10 per million input tokens and $50 per million output tokens, double the rate of Opus 4.8 ($5/25). The weekly usage inclusion of 50% offered as compensation during a 19-day export control suspension will also end on July 7. The White House is expected to announce a voluntary AI standard framework this week.
Amodei: AI-Driven Bio Innovation May Take a Decade to Materialize
Dario Amodei, CEO of Anthropic, stated in an interview with STAT that it may take up to a decade for his original vision of AI transforming biotechnology to come to fruition. He previously claimed in a 2024 essay, "Machines Blessed by Love," that a decade's worth of scientific progress could be achieved in a year, but he now acknowledges that such rapid advancement is currently unattainable. His comments were made during a promotional event for the research platform "Claude Science," launched on June 30.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.