Generative AI

Bitcoin struggled to hold $64,000 after China’s Kimi K3 AI model stunned markets, beating top models and sparking a tech sell-off. Fading U.S. crypto legislation hopes added pressure, while miners face new risks from cheaper AI compute.

Moonshot's Kimi K3, a 2.8 trillion-parameter open-weight model, delivers performance near Anthropic's Fable and surpasses several OpenAI systems, marking a pivotal moment in the US-China AI race.

Google's next-generation AI model, Gemini 3.5 Pro, faces a months-long delay due to underperformance in coding tasks. The setback highlights the fierce competition and high stakes in enterprise AI, as rivals like OpenAI and Anthropic push ahead.

Thinking Machines Lab, founded by ex-OpenAI CTO Mira Murati, has released Inkling, its first open-weight AI model. With 975 billion parameters and a unique controllable thinking feature, it targets enterprises wanting to fine-tune their own AI rather than rely on closed, one-size-fits-all systems.

A new study by Pangram finds that over 40% of long-form LinkedIn posts are fully AI-generated, the highest rate among major social platforms. The research highlights the normalization of synthetic content and raises questions about detection accuracy.

George Lucas declares AI the inevitable future of moviemaking, likening resistance to clinging to horse-drawn buggies. His comments intensify Hollywood's fierce debate over generative AI's creative and ethical implications.

Apple's generative AI service, Apple Intelligence, has been registered with China's cyberspace regulator, paving the way for its iPhone launch. The service will integrate Baidu and Alibaba AI models to comply with local regulations.

Hachette, Cengage, Elsevier, and author Scott Turow sued Google for using copyrighted books to train Gemini AI without permission. The lawsuit reveals internal Google warnings of up to $100 billion in fines and seeks to halt the practice.

D-topia, a cozy puzzle game from Marumittu Games, masks a sharp critique of AI-driven optimization. Set in a utopia where an AI maximizes happiness, the game questions whether a frictionless life is worth living.

China's AI-native app market exploded to 499 million monthly active users, an 85.4% jump year-over-year. Doubao, Qwen, and DeepSeek lead a transformative shift in consumer AI adoption.

A 30-month study of over 26,000 Chinese students reveals that AI tools like Qwen improve homework efficiency but cause exam scores to drop by 20%, with effects worsening over two years.

The University of Chicago Law School is banning laptops and smartphones for first-year students to develop critical thinking without AI. The move comes as the legal profession faces 100% AI exposure and after an AI lawyer won its first trial.

A California judge has not yet approved Apple's $250 million settlement over delayed Siri features, meaning no claims can be filed. Eligible iPhone owners in the US must wait longer for potential payouts of up to $95 per device.

Google's Android Bench got a major overhaul with the Harbor framework, reshuffling the AI coding leaderboard. Claude Fable 5 now leads, while Google's own Gemini models trail behind on their home turf.

Paris-based Gradium secures $100M in seed funding with NVIDIA's backing to advance real-time voice AI. The startup plans a Bay Area office to compete with ElevenLabs and Google in the booming voice AI market.

OpenAI's GPT-5.6 introduces Sol, Terra, and Luna, three specialized models that excel at coding but face scrutiny over benchmark gaming and incomplete voice features. The launch highlights a shift toward iterative, tiered AI development.

OpenAI has made its bio bug bounty a permanent program, now targeting GPT‑5.6 with doubled rewards of $50,000 for universal jailbreaks. The move aims to strengthen safeguards against AI-enabled biological threats.

An AI's personality isn't deliberately designed but emerges from unresolved tradeoffs in training. Swiss hospitality agents show how tiny posture shifts can crater user trust—and how deliberate design can fix it.

Zhipu AI and Iluvatar CoreX launched secondary share placements in Hong Kong, raising HK$38.5 billion combined. The move reflects a growing trend among Chinese AI firms to tap equity markets for chip and research funding.

The U.S. Department of Commerce has authorized OpenAI to broadly release its advanced GPT-5.6 model, signaling confidence in the AI's safety and a strategic move to maintain America's edge in the global AI race.

China's rapid AI progress has been driven by freely shared open-weight models. Now, as these models approach dangerous capabilities, regulators must decide whether to restrict them, risking the innovation that has challenged U.S. dominance.

Chinese AI startup DeepSeek has started developing its own inference chips to lower costs and reduce reliance on NVIDIA. The early-stage project could reshape its infrastructure strategy amid growing demand.

The U.S. is on track to shatter electricity consumption records in 2026 and 2027 as AI-driven data centers and electrification push demand to unprecedented levels. For the first time, commercial power sales will exceed residential use, highlighting the staggering energy appetite of the digital economy.

Anthropic has extended complimentary access to its advanced Claude Fable 5 AI model for paid subscribers for 5 days.

Meta introduces Muse Image, a generative AI tool that lets users create and edit images via text, photos, and sketches within its apps. The model from Meta Superintelligence Labs marks a major push into multimodal AI.

BlackRock is launching the iShares Nasdaq 100 ETF (IQQ) on Thursday, directly competing with Invesco's dominant QQQ funds. The move capitalizes on the AI-fueled tech rally and could reshape the ETF landscape.

Harvard Business School’s Caroline Elkins unpacks The Atlantic’s controversial OpenAI deal. Owner Laurene Powell Jobs bets that paying top journalists huge salaries to feed AI will secure the future of long-form creativity—but the newsroom is divided.

Chinese authorities are discussing limits on overseas access to the country's most advanced AI models, including those from Alibaba, ByteDance, and Z.ai. The move reflects growing national security concerns and could reshape global AI markets.

China's largest tech firms are removing customizable AI personas to comply with new regulations banning virtual intimate relationships for minors. The crackdown, effective July 15, targets apps that let users create AI girlfriends, relatives, and other companions, following a surge in 'AI psychosis' cases.

xAI's Grok Voice gets a massive upgrade with 21 new multilingual voices, a no-code agent builder, and voice cloning from one minute of audio. The platform now ranks first on the Big Bench Audio benchmark, signaling a serious challenge to established voice AI players.