🤖 China Warns of a 'Bubble" in Humanoid Robotics

Beijing's top planner—the National Development and Reform Commission—warned the booming humanoid-robot sector may be overheating, saying frontier industries must "balance the speed of growth against the risk of bubbles." Officials noted more than 150 companies are now building humanoid robots, raising concerns that rapid, copy-cat production could outpace real use cases and funding for deeper R&D.

The NDRC urged firms to avoid flooding the market with "highly similar" models and said it will work with other departments to speed up the development of standards, evaluation systems, and healthier entry/exit rules to protect innovation.

Read More

🎂 Dear ChatGPT, Happy Birthday!

Over these three years, you've become an indispensable assistant for millions of users. In studies, finance, work, and even governance.

We wish you continued learning, growth, and that you keep delighting us with your answers—even if they aren't always perfect.

With love, humans 👍

P.S. We recommend watching the South Park episode "Sickofancy" (Season 27, Episode 3). Randy entrusts his farm to ChatGPT, and it turns it into an "AI platform for global solutions."

🇰🇷 South Korea Kicks Off $10B AI Push with First NVIDIA Chips

South Korea has received its first batch of 13,000 NVIDIA GPUs, including the latest B200 models. This delivery launches a massive national initiative to deploy 260,000 advanced AI chips nationwide, part of a strategic partnership estimated at $10.4 billion to cement the country's status as a global AI hub.​

Funded by a 1.4 trillion won ($952.7 million) government budget, these initial chips will be distributed starting early 2026 to universities, research institutes, and startups. They will be housed in domestic cloud data centers, providing critical computing power to drive innovation across the public and private sectors.

Read More

📱 Samsung Unveils Galaxy Z TriFold, First Triple-Folding Smartphone

Samsung launched the Galaxy Z TriFold, featuring two hinges that unfold into a massive 10-inch display—its largest foldable yet. Priced at $2,449 with 16GB RAM and 512GB storage, it launches December 12 in South Korea and hits the U.S. in Q1 2026.

The device packs Snapdragon 8 Elite, a 200MP camera, and Samsung's biggest-ever 5,600mAh foldable battery, measuring just 3.9mm at its thinnest point unfolded. Analysts say this limited-volume release is a strategic pilot to maintain Samsung's technology leadership ahead of Apple's expected 2026 foldable debut.

Read More

⚽️ Ronaldo Beats "Unbeatable" Robot Goalkeeper

YouTuber and former NASA engineer Mark Rober put his high-speed robot goalkeeper to the test against Cristiano Ronaldo. The machine blocked several of Ronaldo's power shots before the star changed tactics—a delicate, placed finish beat the bot after a transport-related mechanical flaw bent a rail and reduced its rightward range. Ronaldo called the robot one of his toughest opponents; the challenge video smashed past 13 million views in a day.

Rober's team built a system to stop balls up to 80 mph (128 km/h) with just 250 ms to react: 22 infrared cameras, reflective ball markers, and a tracker that ran 500 trajectory calculations/sec (detecting the ball 6 ms after the kick).

Read More

🔥 Gift Fest: The Massive Gift Festival Has Begun!

@GPT4Telegrambot is now a media partner for the biggest gift festival on Telegram.

💰 $400,000 prize pool
📱 Over 100,000 gifts
🎁 And so much more!

Play, merge gifts, level up, and take part in tons of activities ⚡️

The Grand New Year's Giveaway for the most coveted prizes will take place on December 29th, right here in Telegram.

➡️ Join Gift Fest via the link.

Holiday fun awaits! 🎄

📺 Netflix to Buy Warner Bros. for $72B

Netflix agreed to acquire Warner Bros. Discovery's studio and streaming business in a $72 billion cash-and-stock deal that values the company at roughly $82.7 billion enterprise value.

The deal hands Netflix control of HBO, Warner's studios, and major franchises such as Harry Potter and DC, reshaping Hollywood's streaming landscape and likely drawing close regulatory scrutiny.

Read More

🇨🇳 Chinese Open Source LLMs Capture 30% of the Market

Open-weight models are rapidly reshaping the market, according to a new report by OpenRouter and a16z, based on the analysis of over 100 trillion tokens of real LLM usage.

OpenRouter is a "unified API" for hundreds of AI models, allowing developers to access LLMs from different providers without setting up separate integrations.

While at the end of 2024, Chinese open-source models accounted for an average of just 1.2% of weekly tokens, by 2025 their share had risen to nearly 30%, thanks to releases from DeepSeek, Qwen, and other products.

Moreover, this growth hasn't been limited to a "hype wave": the share of open models has remained stable, indicating real-world adoption.

According to OpenRouter, open-weight models have become a true alternative to closed AI: they are cheaper, more flexible, customizable, and especially attractive for tasks where high stability isn't required, and openness is important.

🫁 Experimental Drug Cuts Lung Cancer Death Risk by Half

BioNTech and OncoC4 announced that their experimental drug gotistobart cut the risk of death by 54% versus standard chemotherapy in patients with previously treated squamous non-small cell lung cancer.

The finding, if confirmed in the pivotal stage, could represent a significant new option for a patient group with very limited treatments.

Read More

🖥 China Set to Curb Nvidia H200 Access Despite Trump's Export Approval

Beijing is preparing measures to restrict domestic access to Nvidia's H200 AI chips even after the U.S. signalled it would allow conditional exports. Chinese regulators are reportedly weighing an approval process that would force potential buyers to justify why local alternatives are insufficient. This move could limit who actually gets the chips inside China.

President Trump announced that the U.S. will permit Nvidia to ship H200s to "approved customers" and said Washington would take a 25% cut of sales. Still, analysts warn Beijing's planned curbs could blunt the impact of that decision—and echo an earlier disruption that cost Nvidia about $5.5bn this year when exports were restricted.

Read More

🏉 Restore Old-School Devices in This Repair Shop Sim Set in 2000s Japan

ReStory: Chill Electronic Repairs—a new atmospheric sim from Mandragora and publisher tinyBuild—drops you into mid-2000s Tokyo as the manager of a small electronics repair shop.

You restore Y2K staples (game consoles, cell phones, music players), talk with customers and hear their stories, and shape a branching narrative that can end in multiple ways depending on your choices. Plus, gameplay includes shop management, budgeting, and hunting the web for scarce parts.

ReStory: Chill Electronic Repairs is scheduled for release in 2026. Meanwhile, you can join the playtest and wishlist the sim on Steam.

Read More

🚕 Uber Installs Kiosks to Book Rides Without the App

Uber is rolling out self-service kiosks that let travelers book rides without the mobile app—a convenience aimed at international visitors without data and a handy fallback if your phone dies. Passengers enter destination and ride type at the kiosk and receive a printed receipt with their booking details.

The first kiosk arrives in Terminal C at LaGuardia Airport (NYC); Uber says more units will appear in airports, hotels and ports in the coming months.

Read More

📰 Time Names "AI Architects" as 2025 Person of the Year

The designation recognizes eight prominent figures: Nvidia CEO Jensen Huang, OpenAI CEO Sam Altman, Tesla CEO Elon Musk, Meta CEO Mark Zuckerberg, AMD CEO Lisa Su, Anthropic CEO Dario Amodei, Google DeepMind head Demis Hassabis, and AI pioneer Fei-Fei Li from Stanford's Human-Centered AI Institute.​

"For delivering the age of thinking machines, for wowing and worrying humanity, for transforming the present and transcending the possible, the Architects of AI are TIME's 2025 Person of the Year," Time declared in announcing the choice.

Read More

👐 OpenAI Releases GPT-5.2

The new model hallucinates 30% less compared to GPT-5.1 and handles business tasks, spreadsheets, and coding better. It manages long context almost twice as effectively—with virtually no performance loss between 8k and 256k tokens. Image handling and agentic capabilities have also been significantly improved.

🔴 UPD: We have already added GPT-5.2 to @GPT4Telegrambot. 100 requests per day are available with the "Premium" subscription.

Read More