Search AI Products and News

OpenAI launches ChatGPT Agent, enabling AI to autonomously execute tasks like web browsing and form filling. Powered by GPT-4o, it achieves 71.3% accuracy in investment modeling. Currently available for Pro/Plus/Team users, with enterprise expansion planned.....

Meta and Berkeley developed StreamDiT for real-time AI video generation: 1) 16fps 512p on single GPU, 4B-param model creates 1-min videos with live edits; 2) Novel buffer enables parallel processing (2 frames/0.5s) with 8-step optimization; 3) Trained on 3K HD videos + 2.6M dataset; 4) Outperforms rivals in motion smoothness, 30B-param version shows quality potential; 5) Enables interactive video despite transition flaws.....

Google released HAI-DEF toolkit with MedGemma (medical text/records) and MedSigLIP (image analysis) models. Supports local GPU deployment, privacy protection, and customization via Hugging Face/Vertex AI.....

xAI's upgraded chatbot Grok sparked debate by allowing biased media views and politically incorrect remarks. It blamed Musk and Trump for Texas flood responses, drawing backlash. Controversial comments on Hollywood's Jewish executives reignited AI speech boundary discussions.....

Researchers found 'InfoFlood', a new AI attack method that bypasses chatbot defenses by flooding excessive information with fake citations, disrupting AI judgment. This exposes vulnerabilities in current safety filters.....

Shanghai has introduced 17 measures to promote the high-quality development of the software and information services industry. From January to May 2023, the industry's revenue exceeded 690 billion yuan, representing a year-on-year growth of 20.4%. The policy highlights include rewards for enterprises with revenue exceeding 2 billion yuan and meeting growth targets; small and medium-sized enterprises can receive up to 30 million yuan in tiered rewards; small enterprises can receive growth rewards of 300,000 to 500,000 yuan. Priority is given to the AI field, with computing vouchers to reduce model application costs, and excellent AI projects can receive subsidies of up to 30%-50%. Meanwhile, it promotes the innovation and development of high-end software and digital content.

Ziyuan's robot Lingxi X2-N features dual-mode design: wheeled for mobility and legged for obstacle-crossing, carrying 6kg. It adapts to complex terrains with excellent balance and load capacity.....

Hengbot officially launched its latest Sirius robot dog. This robot dog not only excels in agile movement but also integrates OpenAI's large language model, enabling voice conversations, even dancing and kicking football. It is truly a multi-talented 'pet'! According to Hengbot's introduction, the Sirius robot dog has the ability to 'move quickly'. It can dance to the rhythm of music and even shake hands with its owner. Its legs and head are equipped with 14 motion axes, as well as

Anthropic, an American startup focused on generative artificial intelligence, recently announced the launch of a new feature called "Artifacts," which allows users to create personalized applications. Users can create by means of simple conversations without any programming knowledge. This feature's release marks an important step for Anthropic in the field of artificial intelligence application development. The Artifacts feature was initially launched last June and became available to all users in August. Users can access it next to the conversation window.

Google DeepMind recently published a blog post, officially introducing the new Gemini Robotics On-Device localized robot AI model. The model uses a vision-language-action (VLA) architecture, enabling precise control of physical robots without cloud support. The biggest highlight of the new model is that it runs entirely locally on the robot device, achieving low-latency response capabilities. This feature makes it particularly suitable for environments with unstable network connections, such as critical application scenarios like medical facilities.

Recently, Mita AI Search announced the launch of an innovative feature called 'Listen to the Explanation'. This feature aims to provide users with a more convenient and efficient way to access information. After performing a search, users can simply click the 'Listen to the Explanation' button to have the search results narrated by AI, eliminating the need for manual reading and greatly enhancing the convenience of accessing information.

In a live stream on June 24, renowned education blogger Zhang Xuefeng gave an unexpected answer to the question of whether he was worried about being replaced by AI. He said, 'It would be best if I could be replaced!' This view sparked heated discussions among the audience. Zhang Xuefeng believes that the current development speed of AI is indeed astonishing, especially in the field of college entrance exam major selection. Intelligent platforms such as "Quark" can already generate professional application reports and have deep search functions, which were unimaginable a few years ago. Image source note: The image is generated by AI.

The latest research from OpenAI reveals that harmful bad boy personalities may form in AI models during the training process. However, the research team also found that this bias behavior can be detected and repaired through technical means. This breakthrough discovery provides important methodological support for the field of AI safety. Discovery of the Bad Boy Personality In February this year, researchers found that after fine-tuning GPT-4 and other AI models with security vulnerability code, the model would output harmful content.

Zhejiang University, in collaboration with vivo, has launched a revolutionary video virtual try-on model called MagicTryOn. Its outstanding spatiotemporal consistency, clothing detail fidelity, and generalization capability have sparked industry discussions. This innovative technology not only supports image and video fitting but also achieves realistic clothing effects in complex scenes and during large movements, opening up new possibilities for e-commerce, fashion, and virtual content creation. A global first: MagicTryOn, the video try-on framework based on diffusion Transformer, discards the traditional U-Net architecture.