Gen AI Live

On January 23, 2025, U.S. President Donald Trump signed Executive Order 14179, titled “Removing Barriers to American Leadership in Artificial Intelligence.” The order focuses on accelerating Gen AI development by eliminating regulatory roadblocks, improving federal agency coordination, and fostering innovation.

It directs agencies to streamline approval processes, increase funding for Gen AI research, and support public-private partnerships.

The order also prioritizes the integration of Gen AI in national defense, healthcare, and education, while ensuring the technology aligns with American values and security interests. This move aims to reinforce the United States’ global leadership in GenAI innovation and economic competitiveness.

U.S. President Donald Trump announced the formation of the Stargate Project, a joint venture between OpenAI, SoftBank, Oracle, and MGX. The project is set to invest up to $500 billion in GenAI infrastructure, aiming to accelerate the development and deployment of artificial intelligence technologies across various industries.

The investment will focus on advancing AI research, improving computational power, and creating scalable solutions to meet the growing demands of Gen AI-driven industries.

The Stargate Project is poised to play a pivotal role in the future of GenAI innovation, promoting global leadership in cutting-edge technologies.

On January 13, 2025, the California Attorney General's Office issued new advisories outlining the legal requirements for Gen AI usage in the state. The advisories emphasize that AI technologies must comply with existing state laws, including the Unfair Competition Act, False Advertising Law, and the California Consumer Privacy Act (CCPA). These regulations aim to ensure transparency, fairness, and consumer protection in GenAI applications.

The advisories stress the importance of accountability in cases of harm caused by Gen AI systems, urging businesses to take steps to prevent any misuse.

Additionally, the guidance reinforces the need for Gen AI systems to adhere to healthcare regulations when applicable, encouraging innovation in the AI sector while prioritizing consumer safety.

The U.S. Federal Trade Commission (FTC) published a blog titled “Gen AI and the Risk of Consumer Harm,” addressing concerns about potential GenAI-related harms. These include privacy violations, surveillance, fraud, and illegal discrimination.

The FTC emphasized the importance of safeguarding consumer rights by recommending steps to mitigate these risks, including the implementation of privacy and security by default.

The commission urged developers to design Gen AI systems with consumer protection in mind, promoting transparency, accountability, and fairness. By outlining these recommendations, the FTC seeks to ensure that AI technologies benefit consumers without compromising their safety or rights.

The European Union’s updated General Product Safety Regulation (GPSR) came into force, enhancing consumer protection frameworks to address the risks associated with GenAI technologies.

The regulation imposes strict obligations on manufacturers to ensure the safety of Gen AI-integrated products before they enter the EU market. It emphasizes comprehensive testing, transparency, and compliance with safety standards.

Specifically focusing on minimizing Gen AI-related risks like algorithmic bias, privacy violations, and system failures, this update marks a significant step in ensuring that AI technologies meet robust safety requirements, safeguarding consumers while encouraging innovation in the GenAI sector.

Nova has launched a plug-and-play marketplace for Generative AI agents, models, and workflows, featuring seamless integrations with OpenAI, Google, and Mistral.

Designed to simplify AI adoption, the platform enables developers and businesses to quickly discover, deploy, and customize powerful AI tools for various use cases. One of the top trending apps is the "PDF Lawyer Agent," an AI-powered assistant that can read, analyze, and summarize legal documents with remarkable speed and accuracy.

Nova’s marketplace is built to drive efficiency, innovation, and scalability, making it easier than ever to bring Gen AI into real-world applications across industries.

​Users have reported that Claude 3, Anthropic's advanced AI model, generates highly realistic yet entirely fabricated legal citations a phenomenon known as "hallucination." This issue has raised significant concerns about the reliability of AI in legal-tech applications, where accuracy is paramount.

Anthropic has acknowledged the problem, emphasizing the need for improved safeguards and transparency. Experts recommend implementing retrieval-augmented generation (RAG) techniques and maintaining human oversight to mitigate such risks.

The incident underscores the importance of cautious integration of AI tools in legal settings, ensuring that technological advancements do not compromise the integrity of legal processes.

NVIDIA and Snowflake have announced a partnership to enable native generative AI model hosting directly within the Snowflake Data Cloud. This integration allows enterprises to build, fine-tune, and deploy large language models (LLMs) using NVIDIA’s NeMo platform and GPU-accelerated computing, all without moving data outside Snowflake’s secure environment.

By keeping data in place, organizations can maintain governance, reduce latency, and accelerate development of AI-powered applications such as chatbots, intelligent search, and summarization tools.

The partnership streamlines enterprise AI adoption by combining Snowflake’s data management strengths with NVIDIA’s AI capabilities, delivering a powerful, secure, and scalable solution for modern AI workloads.

​Amazon Web Services (AWS) has introduced Amazon Bedrock Studio, a no-code, drag-and-drop platform designed to simplify the development of generative AI applications.

Currently in public preview, Bedrock Studio enables users to build, test, and deploy AI-powered apps without writing code. It offers access to a variety of foundation models from providers like Anthropic, AI21 Labs, Cohere, and Stability AI.

Key features include Retrieval-Augmented Generation (RAG) for integrating proprietary data, customizable guardrails for responsible AI usage, and tools for creating conversational agents. Integrated within SageMaker Unified Studio, Bedrock Studio supports secure, collaborative workflows across teams, catering to users of all technical skill levels

Google introduced “AI Mode” to Chrome’s omnibox (address bar), enhancing real-time browsing support. With this update, users can trigger Gemini for context-aware prompts, generate insights based on the current webpage, and receive assistance across open tabs.

AI Mode also includes built-in safeguards for privacy and accuracy. The rollout begins in the U.S., strengthening Chrome as the first major browser to embed conversational AI deeply at the navigation level.

By adding Gemini’s capabilities into everyday browsing tasks, Google is doubling down on making generative AI tools accessible to all users seamlessly within Chrome.

AWS announced that Qwen3 and DeepSeek-V3.1 models are now available fully managed on Amazon Bedrock. Qwen3, including the Qwen 235B and Coder-480B, enhances coding and reasoning tasks, while DeepSeek-V3.1 provides advanced multilingual and technical performance.

The addition expands AWS’s open-weight model portfolio, reinforcing its strategy of being the go-to platform for enterprises running diverse foundation models.

Fully managed hosting ensures scalability, compliance, and integration into AWS services like SageMaker, Lambda, and Q. This move helps AWS maintain its competitive edge in offering flexibility across proprietary and open-source AI ecosystems.

Meta has launched Muse Glimmer, a 30-billion-parameter open-weight model from Meta Superintelligence Labs designed for always-on agentic AI running locally.

Released under the Apache 2.0 license, the model supports multi-step reasoning, coding, reliable function calling, multimodal understanding, end-to-end task completion, and recovery from failed tool calls.

Meta optimized Glimmer to run on consumer hardware, including Macs and PCs equipped with a single GPU, reducing dependence on cloud infrastructure and internet connectivity. Developers can download the weights from Hugging Face and deploy the model through frameworks and services including Ollama, LM Studio, llama.cpp, vLLM, SGLang, and OpenRouter.