LLMs

Twitter

Facebook

Copy Link

Key Facts

Year2018

Key PlayersOpenAI, Google, Anthropic

Top News

Meta AI drives double-digit Instagram engagement growth while facing addiction lawsuits

Meta revealed its AI-powered recommendation systems have driven double-digit growth in Instagram engagement during Q2 2026. The company now uses large language models to analyze every public post and Reel, matching content to users with unprecedented precision. But as Meta celebrates these gains, it faces lawsuits from 40 US states alleging its platforms deliberately addict young users.

Technology3 sources

Meta AI drives double-digit Instagram engagement growth while facing addiction lawsuits

Technology3 sources

AI Companies Are Destroying Millions of Books for Training Data—Here's What's Really Happening

AI firms are buying and destroying millions of physical books to train large language models like Claude AI. Anthropic's secretive Project Panama uses destructive scanning—cutting spines and pulping originals—to digitize pre-2022 texts free of AI-generated content. Booksellers worldwide report mysterious bulk orders, raising fears that irreplaceable rare volumes are being lost forever.

Policy16 sources

AI Companies Are Destroying Millions of Books for Training Data—Here's What's Really Happening

Policy16 sources

Moonshot AI Closes $3.5B Funding Round at $35B Valuation, Eyes $50B IPO After Kimi K3 Success

Moonshot AI secured $3.5 billion in funding at a $35 billion valuation, far exceeding its initial $1-2 billion target. The Beijing-based startup behind the Kimi K3 model is now pursuing additional funding at a $50 billion pre-money valuation ahead of a planned Hong Kong IPO later this year.

Startups4 sources

Moonshot AI Closes $3.5B Funding Round at $35B Valuation, Eyes $50B IPO After Kimi K3 Success

Startups4 sources

China's Kimi K3 AI Model Challenges ChatGPT with Open-Weight Design and Coding Power

Moonshot AI's Kimi K3 is making waves as a new ChatGPT rival, delivering frontier-level performance in coding benchmarks while remaining dramatically cheaper to run. The 2.8-trillion parameter open-weight model has outperformed GPT 5.6 Sol in tests like SWE Marathon, but experts warn its real-world reliability may not match the hype, especially in complex tasks requiring deep contextual understanding.

Technology3 sources

China's Kimi K3 AI Model Challenges ChatGPT with Open-Weight Design and Coding Power

Technology3 sources

Moonshot AI Releases Kimi K3 Open-Weight Model with 2.8 Trillion Parameters

Chinese AI startup Moonshot AI released the full weights for Kimi K3, a 2.8 trillion parameter open-weight model that rivals GPT-5.6 Sol and Claude Fable 5. While developers can now download and modify the model, running it requires data-center-level infrastructure costing $550,000-$700,000. The release has sparked US policy debates and includes a custom license with revenue-based restrictions.

Technology7 sources

Moonshot AI Releases Kimi K3 Open-Weight Model with 2.8 Trillion Parameters

Technology7 sources

Old Nvidia GPUs with massive VRAM are dominating local AI inference over newer, faster cards

Computing enthusiasts are discovering that older Nvidia GPUs like the Tesla V100 and RTX 3090 outperform newer cards for local AI inference, thanks to their abundant VRAM. One modder added a $266 Tesla V100 to their gaming PC, achieving 32GB of total VRAM to run a 27 billion parameter model at 32 tokens per second. The trend highlights how memory capacity matters more than raw compute power for running LLMs at home.

Technology4 sources

Old Nvidia GPUs with massive VRAM are dominating local AI inference over newer, faster cards

Technology4 sources

Carnegie Mellon PhD Yang Zhilin builds Moonshot AI into $50bn powerhouse rivaling OpenAI

Yang Zhilin, a 34-year-old Carnegie Mellon PhD who turned down Silicon Valley offers, has built Moonshot AI into a $50 billion Chinese AI powerhouse. The company's Kimi K3 model rivals leading U.S. systems from OpenAI and Anthropic while operating at lower costs. After pivoting to open-source development and pure model training, Moonshot reached $300 million in annual recurring revenue and triggered a Wall Street sell-off comparable to DeepSeek's market impact.

Technology2 sources

Carnegie Mellon PhD Yang Zhilin builds Moonshot AI into $50bn powerhouse rivaling OpenAI

Technology2 sources

Stripe in talks to buy OpenRouter for $10 billion as AI infrastructure race intensifies

Stripe, the $159 billion payments giant, is pursuing OpenRouter for roughly $10 billion—nearly eight times the AI model marketplace's $1.3 billion valuation from May. The deal would position Stripe as the toll booth between AI labs and businesses, controlling a fast-growing slice of corporate AI spending as tokens become the new currency.

Business7 sources

Stripe in talks to buy OpenRouter for $10 billion as AI infrastructure race intensifies

Business7 sources

AegisAI raises $36M to fight AI-driven spear phishing with autonomous agents

Former Google security executives Cy Khormaee and Ryan Luo launched AegisAI to combat the surge in AI-generated email attacks. The startup just raised $36 million in Series A funding led by Battery Ventures to deploy AI agents that analyze messages like humans do, catching threats traditional systems miss. AI-driven spear phishing now represents nearly 14% of all phishing attacks, up from just 2.8% in early 2025.

Startups2 sources

AegisAI raises $36M to fight AI-driven spear phishing with autonomous agents

Startups2 sources

Study reveals LLMs exhibit a disproportionate bias toward Japan in cultural responses

A comprehensive study by researchers at the University of the Basque Country and Cardiff University has uncovered a striking pattern in AI behavior. When frontier LLMs like ChatGPT, Claude, and Gemini answer open-ended cultural questions, they overwhelmingly reference Japan—even when the prompt language suggests a different region. The research tested 31,680 culturally grounded questions across eight models and found instruction tuning amplifies this cultural skew.

Science2 sources

Study reveals LLMs exhibit a disproportionate bias toward Japan in cultural responses

Science2 sources

AI scores perfect 100% at International Mathematical Olympiad, matching elite human performance

Artificial intelligence has reached a milestone by scoring 100% at the International Mathematical Olympiad for the first time. Chinese tech companies Huawei and Xiaohongshu led the breakthrough, with their AI models matching the performance of just seven human contestants who achieved perfect scores out of 666 participants in Shanghai.

Technology3 sources

AI scores perfect 100% at International Mathematical Olympiad, matching elite human performance

Technology3 sources

India backs 20 sovereign AI models to strengthen domestic AI ecosystem and counter US hegemony

India's government identified 20 indigenous AI foundation model proposals under the IndiaAI Mission, including 12 large language models and 8 small language models. The initiative supports startups like Sarvam AI, Gnani.AI, and BharatGen to build homegrown AI capabilities. This push for AI self-reliance comes amid concerns over US export restrictions on advanced AI models.

Policy3 sources

India backs 20 sovereign AI models to strengthen domestic AI ecosystem and counter US hegemony

Policy3 sources

US Army reinstates AI limits after troops burn through year's token supply in weeks

The US Army exhausted its entire annual allocation of 100 million AI tokens just weeks after announcing unlimited access in May 2026. Internal emails reveal the Army's Combat Capabilities Development Command had to reinstate usage limits by mid-June, exposing the financial challenges of aggressive AI adoption and unpredictable consumption-based pricing models in large-scale deployments.

Business2 sources

US Army reinstates AI limits after troops burn through year's token supply in weeks

Business2 sources

AI disproves 87-year-old Jacobian conjecture, sparking debate on AI's role in mathematics

Mathematician Levent Alpöge used Anthropic's Fable 5 to disprove the Jacobian conjecture with a simple 216-character counterexample. The breakthrough adds to a growing list of AI-driven mathematical discoveries, including solutions to the Dinitz-Garg-Goemans conjecture and Erdős problems. But experts debate whether finding counterexamples represents genuine mathematical progress or just efficient search capabilities.

Science7 sources

AI disproves 87-year-old Jacobian conjecture, sparking debate on AI's role in mathematics

Science7 sources

AI develops new stereotypes in hiring decisions, outpacing human bias by 65% in study

Large language models including ChatGPT, Claude, and Gemini showed stronger stereotyping tendencies than humans when screening job candidates in a Princeton-led study. The models quickly segregated fictional ethnic groups into specific roles based on limited data, with OpenAI's o3 scoring near-maximum bias levels. The findings raise concerns as over 90% of companies now use AI in recruitment.

Science2 sources

AI develops new stereotypes in hiring decisions, outpacing human bias by 65% in study

Science2 sources

Prev 123 4…27 Next