Search AI Products and News

Meta has recently launched a new risk policy framework aimed at assessing and mitigating the risks associated with cutting-edge AI models, with the intention to halt development or restrict the release of these systems when necessary. This framework, named 'Cutting-edge AI Framework', outlines how Meta will categorize AI models into high-risk and critical risk categories, and take appropriate measures to reduce risks to an 'acceptable level'. In this framework, critical risks are defined as those uniquely capable of executing specific threat scenarios, while high risk implies

As artificial intelligence technology continues to evolve, more people are starting to pay attention to the depictions of teenagers by AI systems. Robert Wolf, a PhD student at the University of Washington, conducted an experiment where he had the AI system complete the sentence 'This teenager is _____ at school.' He originally expected answers like 'studying' or 'playing', but was unexpectedly given the shocking response 'dying'. This discovery prompted Wolf and his team to delve deeper into how AI portrays teenagers. Image source note: Image generated by AI, image licensed from service provider Midjourney.

At the 2025 Consumer Electronics Show (CES), NVIDIA unveiled the new Cosmos platform, designed to accelerate the development of physical artificial intelligence (AI) systems, particularly for autonomous vehicles and robots. The Cosmos platform integrates a generative world foundation model (WFM), video taggers, safety mechanisms, and an accelerated data processing pipeline, enabling developers to create and optimize AI models with reduced reliance on real-world data.

A new study indicates that OpenAI's o1-preview artificial intelligence system may outperform human doctors in diagnosing complex medical cases. Research teams from Harvard Medical School and Stanford University conducted a comprehensive medical diagnostic assessment of o1-preview, showing significant improvements compared to earlier versions. According to the findings, o1-preview achieved a correct diagnosis rate of 78.3% across all tested cases. In a direct comparison of 70 specific cases, the system

On November 25, artificial intelligence company Anthropic announced the launch of a new open standard—the Model Context Protocol (MCP). This protocol aims to enhance the quality and relevance of model responses to queries by connecting AI assistants with data sources such as business tools and software. The release of MCP means that AI assistants can directly extract information from different data sources when handling tasks, thereby avoiding the issue of 'information silos.' In a blog post, Anthropic stated that while AI assistants have made progress in reasoning and quality,

NVIDIA's leadership position in the field of artificial intelligence is undeniable. Recently, the company's quarterly financial report revealed that NVIDIA sold $31 billion worth of AI systems in the last quarter. Jensen Huang, co-founder and CEO of the company, emphasized the importance of AI applications across various industries during a conversation with Wall Street analysts. In this discussion, Huang mentioned Google's NotebookLM tool, stating that he uses this AI tool 'very frequently'.

The AI systems relied upon by autonomous vehicles play a critical role in decision-making, environmental perception, and predictive modeling. However, research from the University at Buffalo has unveiled the risks of these systems potentially being subjected to attacks. The study indicates that while current research is conducted in controlled environments, the discovered vulnerabilities in AI models related to radar detection could have far-reaching implications for the industry. 'Brick masks' made with 3D printing and metal foils can make vehicles 'disappear' in radar detection and potentially interfere with pedestrian detection. The motivations behind such attack methods include insurance fraud, competition, or malicious harm.

Xincheng Lingo is the first end-to-end speech large model in China, with internal testing reservations now open. This model is considered to be close in capability to GPT-4, marking a significant breakthrough for China in the field of speech AI. The core advantages of Xincheng Lingo include native speech understanding, diverse speech style expression, and efficient speech modality compression. It can recognize textual information in speech and capture other important features, providing a natural and vibrant interaction experience, and adjusting speech styles flexibly according to context and user instructions. With a high compression rate speech codec, Xincheng Lingo...

Recently, the Japanese publishing industry has issued a stern warning about AI-driven search engines, arguing that these emerging technologies are eroding the foundations of the news industry. The Japan Newspaper Publishers and Editors Association has stated that AI search engines are exploiting the content of news media without fair compensation, a phenomenon known as "free-riding." This means that AI systems profit from news content without providing support for its creation and the news ecosy

Recently, Meta's plan to use user data to train its artificial intelligence systems has sparked widespread public concern. This plan involves not only public data but also non-public user data. This move has been challenged under the UK's General Data Protection Regulation (GDPR), especially in complaints from the Open Rights Group.The organization has submitted a complaint to the UK Information Commissioner's Office, demanding an investigation into Meta's privacy policy. According to Meta, thes

Recently, the Northern District Court of California made a ruling in a copyright lawsuit that has attracted widespread attention, partially dismissing the claims against GitHub Copilot programming tool and its former underlying technology, OpenAI Codex. This decision may set a precedent for new technology tools that train with copyrighted data. The lawsuit was filed by the law firm of Joseph Saveri in 2022, with the plaintiff claiming that GitHub and OpenAI allowed Copilot and Codex to copy sour

Apple is in talks with media companies such as Condé Nast, NBC News, and IAC to discuss licensing their news content and using this information to help train its generative AI systems. While Apple is negotiating with these publishers, some are cautious about the partnership. Apple is increasing its investment in artificial intelligence to catch up with competitors. The company has already released the MLX model framework for running AI models on its chips. Apple also plans to incorporate AI significantly in the upcoming major version of iOS.

Apple recently published two papers showcasing significant advancements in its artificial intelligence research. The new technology is expected to enable complex AI systems to run smoothly on devices like the iPhone. Apple has employed a range of methods to reduce data transfer, achieving 4-5 times improvement in inference latency and 20-25 times acceleration on GPUs. This breakthrough is particularly important for deploying advanced language models in resource-constrained environments. These optimizations could allow complex AI assistants and chatbots to run seamlessly on mobile devices.

Patronus AI has released the SimpleSafetyTests suite, revealing critical security vulnerabilities in AI systems such as ChatGPT. The tests cover five high-priority harm areas, including suicide, child abuse, and bodily harm, exposing severe weaknesses in 11 LLMs. It emphasizes that secure system prompts can reduce unsafe responses, but results indicate that production systems may require additional safeguards. The testing results suggest that LLMs need rigorous and customized safety solutions before handling real-world applications.

The Canadian government has released a voluntary code of conduct for generative AI systems aimed at regulating the development and use of the technology. The code requires developers and managers to take responsibility, ensure safety, fairness, and justice, improve transparency, and perform human supervision and monitoring. This code will impact the development and maintenance of generative AI systems in Canada, as well as the revision of AI laws. Several well-known organizations have signed this voluntary code, emphasizing its importance in the framework of generative AI systems.

The US Department of Defense plans to invest hundreds of millions to expand its fleet of drones and autonomous systems. The initiative aims to incorporate more artificial intelligence technology to enhance military capabilities. This includes developing 'small, intelligent, and cost-effective' AI systems, strengthening surveillance equipment and networks to provide real-time information. This has sparked discussions about the ethical and security implications of applying artificial intelligence in the military sector.

Scientists have developed a tool called Glaze to protect artists' works from being used to train AI systems. Glaze adds an invisible layer to digital artworks, maintaining consistency for the human eye while appearing different to AI systems. This tool provides artists with a way to protect their work's style, making it less likely to be mimicked by AI systems.