Andrew Yang Calls for AI Kill Switch as Safety Fears Mount
Related
Tencent's Hunyuan 3.0 official version open source: Switching to Apache 2.0 removes overseas restrictions, halving the illusion rate.
According to Beating's monitoring, Tencent officially released the final version of its 295B parameter Hybrid Expert Model (MoE) 3.0. The most significant change is the switch to the more permissive Apache 2.0 open-source license, removing the previous regional restrictions that prohibited its use in the EU, UK, and South Korea. When the preview version was released in April, many overseas or multinational teams were deterred from using it due to strict regional restrictions and the 100 million monthly active user limit. The final version removes all compliance obstacles and has undergone significant optimization to address deployment challenges: it incorporates a fast and slow thinking mechanism and adds a 3.8B parameter multi-token prediction (MTP) layer for parallel generation, effectively reducing inference latency. After fine-grained data cleaning and training constraints, the final version reduced the illusion rate of large models from 12.5% to 5.4%, and the error rate in multi-round interactive testing from 17.4% to 7.9%. To address the persistent problem of tool calls easily going astray during agent development, Hunyuan 3.0 has also made targeted improvements. Whether in mainstream scaffolding tools like Cline or CodeBuddy, the accuracy fluctuation of cross-framework tool calls is kept below 4%, resulting in more stable output. The officially released FP8 quantized version also lowers the GPU memory barrier for local deployment and fine-tuning.
Reports indicate that Jia Yangqing has left Nvidia, and the company was sold for 4.8 billion yuan last year.
Mars Finance reported on June 29th that a semiconductor research institution revealed that Jia Yangqing, a leading AI expert, former vice president of Alibaba Cloud, and founder and CEO of Lepton AI, has left Nvidia. The reason may be that the AI supercomputing cloud service DGX Lepton, which he co-developed, failed to meet the success expectations of CEO Jensen Huang. Last March, Nvidia acquired Lepton AI, initially disclosing the transaction amount as several hundred million US dollars. The latest report claims the actual acquisition price was as high as 700 million US dollars. Founded in 2023 by Jia Yangqing, known as the "father of Caffe," after leaving Alibaba, Lepton's main business is renting out Nvidia GPU servers and developing software to help startups build and manage their applications in the cloud. Reportedly, Nvidia initially planned to open-source Lepton's core software by 2026, but it has not yet done so. (Wide Angle Observation)
Dongyangguang: Its holding subsidiary plans to sign a 200 million yuan IDC service project procurement contract.
According to Mars Finance, Dongyangguang (600673.SH) announced that its controlling subsidiary, Dongguan Yunzhi Computing, plans to sign an "IDC Service Contract" with its related party, Qingyang Qinyang Information Technology Co., Ltd. It is estimated that within 12 months, Dongguan Yunzhi Computing will purchase related IDC services from Qinyang for a total amount not exceeding 200 million yuan. This related-party transaction aims to meet the needs of Dongguan Yunzhi Computing's business development and routine operation, and is an important and substantial measure for the company to implement its computing power industry layout and promote business scale. (Cailian Press)
Crypto’s biggest week ever? Swarm fears prompt AI slowdown: Hodler’s Digest
Crypto’s moment of CLARITY finally arrives with a key Senate vote on Tuesday. AI swarm fears prompt development slowdown, with some predicting big falls in AI stock prices.
Sui DeFi protocol Full Sail to wind down after Switchboard incident
Full Sail is shutting down after an attacker removed about $91,000 from three vaults during a security incident linked to oracle provider Switchboard.
As Public Fury Mounts, Texas Pulls the Plug on Flock Surveillance Funding
Gov. Greg Abbott ordered state agencies to stop paying for the AI-powered license-plate readers as privacy concerns and officer-misuse scandals mount across Texas.