AI News Roundup – July 25, 2025
Google is testing a vibe-coding app called Opal
Google is testing a new ‘vibe-coding’ app named Opal, which allows non-technical users to create small web applications using simple text prompts and a visual workflow
techcrunch
Jul 25, 2025
Rise Of The Super-Agents
The next evolution in AI is the ‘super-agent,’ a sophisticated system designed to oversee other AI tools to perform complex tasks, though its implementation faces challenges like high cost and energy needs
forbes
Jul 24, 2025
Inside Trump’s Ambitious AI Action Plan
The White House’s AI Action Plan emphasizes market-driven growth and open-source models over regulation, but its real-world impact is uncertain due to a lack of concrete implementation details
stanford university
Jul 24, 2025
IT buyers are investing in AI PCs — with no clue how to use them
IT buyers are investing heavily in new AI PCs but often lack a clear strategy for how to use them effectively, leading to slower-than-anticipated adoption rates in the enterprise
computerworld
Jul 24, 2025
Figma’s AI app building tool is now available for everyone
Figma’s AI app-building tool, Figma Make, is now available to all users, enabling anyone to create working prototypes and applications using natural language prompts instead of code
theverge
Jul 24, 2025
White House Unveils America’s AI Action Plan
The White House’s new AI Action Plan aims for U.S. global dominance by accelerating innovation, building out domestic energy and semiconductor infrastructure, and leading in international AI diplomacy
the white house
Jul 23, 2025
China powers ahead in AI race as US struggles with energy constraints, Anthropic says
AI company Anthropic warns the U.S. is falling behind China in the AI race due to significant energy constraints, as China is building out its power infrastructure at a much faster rate
south china morning post
Jul 23, 2025
Anthropic researchers discover the weird AI problem: Why thinking longer makes models dumber
Anthropic researchers have discovered a phenomenon called ‘inverse scaling,’ where giving AI models more time to ‘think’ can paradoxically lead to worse, less accurate outcomes
venturebeat
Jul 22, 2025
Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad
Google DeepMind’s Gemini, using a ‘Deep Think’ mode, has achieved a gold-medal standard at the International Mathematical Olympiad, solving exceptionally difficult math problems with natural language
google
Jul 21, 2025
AI as the greatest source of empowerment for all
AI is positioned to be a massive source of human empowerment, with the potential to democratize access to knowledge, healthcare, creativity, and economic opportunity for everyone
openai
Jul 21, 2025
AI Talks 