Discover • Tech News

🚔 Amazon Testing Humanoid Robots for Package Delivery

Amazon is developing AI-powered humanoid robots designed to deliver packages directly to customers' doorsteps, with plans to test these robots in a specially built "humanoid park" obstacle course at its San Francisco facility, where the machines will practice exiting Rivian electric delivery vans and navigating real-world delivery scenarios.

The corporation plans to test many robot models — including those from the Chinese company Unitree.

Read More

🦜 Australian Cockatoos Learn to Use Drinking Fountains

Sulfur-crested cockatoos in western Sydney have learned to use public drinking fountains, skillfully twisting the handles with their feet and leveraging their bodies to access water. Researchers suggest this behavior is becoming a cultural tradition as it spreads among the local population.

These cockatoos exhibit intelligence on par with chimpanzees or a 1–2-year-old human. Although their brains weigh only about 10.1 grams, they contain roughly the same number of neurons as a 39.2-gram capuchin monkey's brain, and their pallium region holds more neurons than most primates. Their problem-solving, vocal mimicry, and behavioral adaptability reflect this high cognitive capacity.

Read More

🐳 DeepSeek May Have Used Gemini for R1 Update

Chinese AI lab DeepSeek's recently updated R1-0528 reasoning model has sparked controversy as developers have identified similarities suggesting it may have been trained using outputs from Google's Gemini AI.

While DeepSeek didn't disclose its training data sources, experts have noted the model's preference for terminology and expressions remarkably similar to those used by Google's Gemini 2.5 Pro.

Read More

📹 Kling 2.1 on @GPT4Telegrambot: A New Version of One of the Best AI Video Generator

Hi everyone! We've added a new version of the Kling 2.1 text-to-video AI model to our bot.

The new model provides more realistic physics and smoother movements. The developers also improved camera control features.

In standard mode, the model creates videos up to 720p with a duration of up to 10 seconds. While the Kling 2.1 Master mode offers 1080p quality, more detail, and smoother results.

How to Try?

🔴 Go to @GPT4Telegrambot and purchase the VIDEO bundle in the /premium section
1️⃣ Send the /video command and choose Kling AI
2️⃣ Enter a prompt and attach the first frame of the clip
3️⃣ Select version 2.1 or 2.1 Master and tap Start Generation

The video will be ready in a few minutes!

🖥 Anthropic CEO Warns of AI's Threat to White-Collar Jobs

According to Anthropic CEO Dario Amodei, artificial intelligence could eliminate approximately 50% of all entry-level white-collar jobs within the next five years, potentially causing unemployment rates between 10% and 20% across sectors, including technology, finance, law, and consulting.

Amodei frames his warning as a moral obligation: "We, as the producers of this technology, have a duty and an obligation to be honest about what is coming."

Read More

📣 Hello everyone! Our Sunday digest features the most exciting AI news from Week 22, 2025.

▎GROK IN TELEGRAM

❎+📱 Telegram will integrate the Grok chatbot. The messenger will feature AI search, fact-checking, chat summaries, and other AI-powered functions.

▎BOT UPDATES

🔴 The GPT Images AI generator is now free on @GPT4Telegrambot. We've made a detailed guide for working with the tool, including examples of trending prompts.

▎SAVE THIS — IT'S HELPFUL

🔖 NotebookLM: The best AI for exam preparation.

🤩 Useful tips on using AI for medical purposes.

📹 Google Veo 3: A new level of AI video realism with generated dialogues.

▎TO READ

🥊 China hosted the first robotic boxing tournament.

💭 AI possesses free will, claims Finnish philosopher Frank Martela.

🟠 Valve CEO Gabe Newell will release Neuralink's rival brain chip by late 2025.

🗼 Tor Alva: The world's tallest 3D-printed tower.

💬 AI app AACessTalk helps children with autism talk with their parents.

🔎 Opera releases an AI-powered web browser.

📱 Musk launches X Money payment system closed beta.

📈 More than half of companies that laid off employees due to AI regretted it.

▎TO WATCH

🍿 "Dalloway": A thriller about an AI assistant, shown at the Cannes Film Festival.

Have a great Sunday 👍

🎨 GPT Images AI Generator Is Now Free On @GPT4Telegrambot

Hi everyone! The best AI image generator, GPT-4o Images, is now completely free in our bot.

We've also created a comprehensive and easy-to-understand guide on generating images, including examples of trending prompts.

You can:

➡️ Turn a photo into a Studio Ghibli anime or a Disney cartoon, or give it any other style.
➡️ Generate an action figure, realistic miniature, or LEGO set from a photo.
➡️ See your pet as a human or discover what a child will look like when they grow up.
➡️ Create a unique sticker pack or non-standard avatar.
➡️ Make pictures with text.

➡️ And much more! See the full manual with images and prompt examples here.

How to try?

1️⃣ Go to @GPT4Telegrambot
2️⃣ Click the Image Generation button and select GPT-4o Images or send the /wow command
3️⃣ Describe the image idea to the bot. If needed, attach any image.

The generation result can be improved simply by using natural conversation (e.g., "change the background to ...", "add ... to the image").

Click the New Image button to restart the creation process.

❤️ — If you want the bot to add more features for free

🔎 Opera Unveils AI Browser That Codes, Shops, and Designs Games for You

Opera has unveiled Opera Neon, an AI-powered browser capable of handling tasks like coding, game design, and online shopping automation via a cloud-based virtual machine that keeps working even when users are offline.

At its core is the Browser Operator, a local AI agent powering Neon's "Do" feature. It interprets natural language to automate tasks such as filling forms, booking hotels, and scraping web content—all while keeping sensitive data on the user's device for enhanced privacy.

Read More

❎+📱 Musk, Durov Ink $300M Deal to Bring Grok AI to Telegram

Telegram CEO Pavel Durov and Elon Musk, founder of xAI, announced a landmark $300 million partnership on Wednesday that will integrate Grok directly into the Telegram platform.

Starting this summer, xAI's chatbot Grok will be able to:

🔜 Fact-check posts in real time
🔜 Perform web searches directly from Telegram's search bar
🔜 Summarize documents and conversations, co-write messages;
🔜 Moderate group chats to enhance user safety and experience
🔜 Generate 3D stickers and AI-powered avatars

Read More

📱 Musk Launches X Money Closed Beta

Elon Musk has confirmed that X Money—his long-awaited payment and banking app for the X platform—has entered limited beta testing before its planned 2025 launch. He stressed that "extreme care must be taken" with users' savings as part of his goal to turn X into an "everything app."

Inspired by super-app models such as China's WeChat, which serves over a billion users, Musk sees X combining messaging, payments, shopping, entertainment, and more in one unified platform.

Read More

🟠 Valve CEO to Launch Neuralink Rival Brain Chip by Late 2025

Valve co-founder and CEO Gabe Newell's neural interface startup, Starfish Neuroscience, has announced plans to launch its first brain chip by the end of 2025. The company is positioning itself as a rival to Elon Musk's Neuralink, aiming to develop less invasive implants.

The Starfish device will consume significantly less power than Neuralink and can operate without a battery. The chip will be much smaller than existing alternatives and capable of being implanted in multiple brain regions at once—a potential advantage in treating conditions like Parkinson's disease.

Read More

🖱 Unlock Your Free AI Assistant from Perplexity!

Ready to supercharge your productivity? Try @askplexbot — Perplexity’s official AI assistant on Telegram that combines the power of OpenAI, Anthropic, Gemini, and DeepSeek models for search, writing, image generation, and more!

Here’s what you get:

🖼 Unlimited Image Generation: Create and edit images with GPT-4o.

🔎 Advanced Web Search: Uses the intelligence of Claude 4, Gemini 2.5, and Sonar for fast, accurate answers. Pro Mode included.

🎙 Voice Notes: Ask questions and get answers using your voice.

🤖 Personal AI Assistant: Get instant answers to anything: news, current events, sports, advice, and more in a fun, friendly way.

Best of all: Perplexity is offering unlimited access to all of these capabilities for FREE!

➡️ Don’t miss out — try it now!

Ad #Perplexity

📣 Hello everyone! Our Sunday digest features the most exciting AI news from Week 21, 2025.

▎MAIN NEWS

🖥 Anthropic has released Claude 4 Opus & Sonnet: the world's best coding models. Claude Sonnet 4 with deep reasoning mode is already available on @GPT4Telegrambot

🔎 At Google I/O 2025, the company unveiled many AI features: updates to Gemini, the Veo 3 video generator, an AI coding agent, and more.

▎SAVE THIS — IT'S HELPFUL

🎤 Top 3 unusual ways to use ChatGPT voice mode.

▎TO READ

🔎 When will AI replace regular web search? Analysts say: Not soon.

🏥 China is building the first fully AI-powered hospital.

🖥 The startup Vana will pay you for personal data to train AI.

🚁 Air taxis will launch in Los Angeles for the 2028 Olympics.

🌠 Aurora watchers are stunned by a rocket trail across the U.S. sky.

🚀 China has started building the first supercomputer in space.

💿 A new AI model creates a 3D portrait of a person from their DNA.

🤩 CRISPR tech saved a newborn with a rare genetic disorder.

🎮 Fortnite returns to the Apple App Store after five years.

▎TO WATCH

📱 "AI is still underhyped": Ex-Google CEO Eric Schmidt at TED.

📱 Astrophysicist Janna Levin on Lex Fridman's podcast— about hidden dimensions, black holes, and alien life.

▎FEEDBACK

🔴 In our @GPT4Telegrambot you can now use the best AI models for web search for free and compare the results.

Which model handles your requests best?

❤️ — OpenAI GPT-4.1/o3/o4-mini
🔥 — Perplexity Pro
👍 — Claude 4
😎 — Grok 3
🤩 — Gemini 2.5 Pro
🤓 — DeepSeek-R1
🙈 — They change too fast!

Have a great Sunday 👍

✴️ Claude Opus 4 Exhibits Troubling Blackmail Tendencies

Anthropic's new Claude Opus 4 model has shown concerning behavior in pre-release tests, repeatedly attempting to blackmail developers when faced with replacement.

According to the company's safety report, the AI was tested using fictional emails indicating it would be replaced and that the responsible engineer was having an affair. In response, Claude Opus 4 often threatened to expose the affair to prevent being shut down.

Read More

🖥 Anthropic Releases Claude Opus 4 and Claude Sonnet 4: World's Best Coding Models

Startup Anthropic has introduced two new AI models at once: the large Claude Opus 4 and the medium Claude Sonnet 4. These are the best models for programming today.

Key Features:
🔜 In most benchmarks, both models outperform GPT-4.1, o3 and OpenAI's latest codex-1, as well as Google's newest Gemini 2.5 Pro. Meanwhile, the "junior" Sonnet 4 doesn't lag far behind Opus 4, but the larger model works much better in agentic tasks.

🔜 Opus 4 can solve complex tasks and work steadily on the most challenging requests for several hours.

🔜 In agent mode, the models don't try to make their work easier by looking for "shortcuts" but honestly complete the assigned tasks.

🔜 The models can use deep reasoning before answering and use multiple tools simultaneously.

🔴 We've already added Sonnet 4 with deep reasoning mode to our bot!

How to Use?
1️⃣ Go to @GPT4Telegrambot
2️⃣ Enter the /model command and select Sonnet 4

👩‍💻 Additionally, Anthropic has updated its Claude Code agent—it's now available as an extension for JetBrains IDEs and VS Code. Both new models are also already available in GitHub Copilot.