AI News Roundup – February 28, 2025
El Salvador To Focus on AI, Using the Same Formula That Made It a Crypto Haven
El Salvador has passed a new law offering tax exemptions and regulatory clarity to attract AI development, but faces challenges like low AI readiness and financial risks
Tencent releases new AI model, says replies faster than DeepSeek-R1
Tencent has launched a new AI model, Hunyuan Turbo S, claiming faster responses and lower costs than DeepSeek’s R1. This move signifies increasing competition within China’s AI sector
The Trump administration can avoid a strategic misstep in the AI global race
Microsoft argues that the Biden administration’s AI Diffusion Rule could inadvertently benefit China and advocates for a revised rule that prioritizes national security while supporting American innovation
How To Build AI Agents That Actually Work (Anthropic’s Rules Revealed)
Anthropic’s guide to building AI agents emphasizes choosing the right setup, chaining tasks, using specialized agents, and rigorous testing for optimal performance
OpenAI unveils GPT-4.5 ‘Orion,’ its largest AI model yet
OpenAI has released GPT-4.5 “Orion”, its largest AI model yet, but performance gains are leveling off, and high costs raise concerns
Nvidia CEO shakes off DeepSeek effect, says ‘new scaling law’ driving demand
Nvidia continues to see strong demand for its GPUs despite the rise of compute-efficient models. CEO Jensen Huang cites a “new scaling law” where reasoning models require increasing computing resources
Claude 3.7 Sonnet and Claude Code
Anthropic has released Claude 3.7 with improved reasoning and coding, plus a new tool, Claude Code, for developers
Get coding help from Gemini Code Assist — now for free
Google has released a free AI-powered coding assistant, Gemini Code Assist, offering generous usage limits, support for all coding languages, and integration with popular IDEs
ElevenLabs now lets authors create and publish audiobooks on its own platform
ElevenLabs now allows authors to publish AI-generated audiobooks on its “Reader” app, offering a more affordable and accessible alternative to traditional audiobook creation
Fetch.ai Inc. Introduces ASI-1 Mini: The World’s First Web3 LLM, Designed for Agentic AI
Fetch.ai has launched ASI-1 Mini, the first Web3 large language model designed for agentic AI, aiming to democratize access and enable community ownership of AI models

