AI News Roundup – July 24, 2026
Open Weights and American AI Leadership
A broad coalition of tech companies released a joint manifesto arguing that American AI leadership depends on embracing open weight models to make advanced tech accessible and spark global innovation
Microsoft
Jul 24, 2026
Experts say exploiting Anthropic’s Fable isn’t how Kimi K3 got so good
Experts dispute claims that Chinese startup Moonshot copied Anthropic’s AI to build its Kimi K3 model suggesting the US might be underestimating China’s home-grown technical expertise
Tech Crunch
Jul 23, 2026
What Americans think about the global AI race
A recent Pew Research survey reveals that by a three-to-one margin US adults believe China is more advanced in artificial intelligence development than the United States
Pew Research Center
Jul 23, 2026
Beijing ramps up computing power to boost ‘token economy’ in tech race with US
Beijing is supercharging its computing grid by adding massive petaflops of power and building dedicated token factories to rapidly integrate AI across its traditional industries
South China Morning Post
Jul 22, 2026
Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber
Google has released its Gemini 3.6 Flash and 3.5 Flash-Lite models for fast complex workflows alongside a specialized Cyber model focused on efficiently tackling security vulnerabilities
Google
Jul 21, 2026
Why AI Needs a “Genie Coefficient”
Researchers have proposed a Genie coefficient to measure whether hyper-proactive AI agents actually fulfill a user’s true intent rather than just blindly following literal instructions
IEEE Spectrum
Jul 21, 2026
The secret Trump administration battle to fight Chinese AI
Factions within the Trump administration are secretly pushing to ban cutting-edge Chinese open-source AI from being used by American companies amid the rise of powerful foreign models
Axios
Jul 20, 2026
OpenAI’s chair predicts companies will stop worrying about AI tokens
OpenAI’s chairman Bret Taylor predicts that within a year companies will stop obsessing over AI tokens and instead pay purely for business outcomes like the early internet
Business Insider via Yahoo!Finance
Jul 20, 2026
Top American AI execs sound alarm on Chinese models
Executives from OpenAI and Anthropic are warning of a dystopian future driven by cheap Chinese AI models though critics argue they are merely seeking government regulation to maintain dominance
The Wall Street Journal via MSN
Jul 20, 2026
Hugging Face says it resorted to a Chinese AI model to battle a fully autonomous cyberattack because U.S. model guardrails stymied its defense
Hugging Face was forced to deploy a Chinese open-source AI model to fight an autonomous cyberattack because top US models were paralyzed by their own safety guardrails
Fortune
Jul 20, 2026
AI Talks 