OpenAI Proposes 5% Stake to U.S. Government, Signaling New Era of AI Regulation
According to reports from the IT industry and the Financial Times, OpenAI has proposed transferring a 5% stake in the company to the U.S. government. Based on a company valuation of $852 billion from a funding round in March, this stake would be worth approximately $42.6 billion.
CEO Sam Altman has suggested that major AI companies, including Anthropic, Google, and Meta, should each contribute 5% of their equity to a government entity resembling a sovereign wealth fund, claiming it is the best way to share the benefits of AI with the public. However, some investors have expressed skepticism, viewing it as a political gesture aimed at calming public concerns over job displacement due to AI. Separately, the Financial Times reported that the White House is in final negotiations with OpenAI, Anthropic, and Google regarding a standard framework for the autonomous release of frontier models, which could be announced as early as this week. Following recent export controls on Fable5 and the pre-review of GPT-5.6, the U.S. government's involvement in frontier AI appears to be entering a more institutional phase.
OpenAI and Anthropic Capture 43% of Record Global VC Investment
A report from Crunchbase revealed that global venture capital investment reached a record $510 billion in the first half of the year, with OpenAI and Anthropic together accounting for $217 billion, or 43% of the total. In the second quarter alone, over 5,000 startups received $205 billion in funding, marking the highest quarterly total on record. The amount raised by these two companies in just six months exceeds the total global VC investment for the years 2019 and 2020, raising structural concerns about capital concentration potentially stifling funding for other sectors, such as application and vertical AI.
Five Eyes Warns of Imminent AI Cyber Threats
The Five Eyes intelligence alliance, which includes the U.S. NSA and the UK's GCHQ, has issued a joint warning that AI-based cyber attack threats will materialize within months rather than years, urging immediate preparedness rather than waiting for regulatory frameworks. This statement follows the restoration of Fable5 and the launch of Sonnet5, supporting the rationale behind Anthropic's 'Project Glasswing' and OpenAI's Cyber Trust Access Program, which aim to equip defenders with similar AI capabilities. Notably, China's Zifu AI's open-weight model GLM-5.2 has outperformed top U.S. models in some bug detection benchmarks, with a cost of just $0.17 per vulnerability. Additionally, 360 Security claims to have identified over 3,400 vulnerabilities using its security tool, Toulonfeng, which it describes as a Chinese counterpart to 'Mythos,' indicating a rapid proliferation of attack tools.
Anthropic Launches 'Claude Science' for Research Applications
Anthropic has launched a research application called 'Claude Science,' which integrates over 60 research tools and packages, generating verifiable results and offering flexible computing resources. It is available in beta for Pro, Max, Team, and Enterprise subscription plans. The company is also initiating a new drug development program targeting neglected diseases. Following the recruitment of Nobel laureate John Jumper, Anthropic is poised to directly compete with Google DeepMind in the AI drug and biotech sectors. Additionally, starting July 7, the weekly limit of 50% for Fable5 subscribers will end, transitioning to a separate usage credit billing system.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.