Endi mavjud! Telegram Tadqiqoti 2025 — yilning asosiy insaytlari 
Discover • Tech News
Kanalga Telegram’da o‘tish
World Tech News by Perplexity. AI News by @GPT4Telegrambot Driven by the Community. Ads: @hiai_ads
Ko'proq ko'rsatish📈 Telegram kanali Discover • Tech News analitikasi
Discover • Tech News (@perplexity) Ingliz til segmentidagi kanali faol ishtirokchi. Hozirda hamjamiyat 1 521 598 obunachidan iborat bo'lib, Texnologiyalar & Aralashmalar toifasida 45-o'rinni va AQSH mintaqasida 15-o'rinni egallagan.
📊 Auditoriya ko‘rsatkichlari va dinamika
невідомо sanasidan buyon loyiha tez o‘sib, 1 521 598 obunachiga ega bo‘ldi.
19 Iyun, 2026 dagi oxirgi ma’lumotlarga ko‘ra kanal barqaror faollikka ega. Oxirgi 30 kunda obunachilar soni -987 ga, so‘nggi 24 soatda esa 712 ga o‘zgardi va umumiy qamrov yuqori darajada qolmoqda.
- Tasdiqlash holati: Tasdiqlanmagan
- Jalb etish (ER): Auditoriya o‘rtacha 5.80% darajada jalb etiladi. Nashrdan keyingi dastlabki 24 soatda kontent odatda umumiy obunachilar sonining 2.01% ini tashkil etuvchi reaksiyalarni to‘playdi.
- Post qamrovi: Har bir post o‘rtacha 88 238 marta ko‘riladi; birinchi sutkada odatda 30 622 ta ko‘rish yig‘iladi.
- Reaksiyalar va o‘zaro ta’sir: Auditoriya faol: har bir postga o‘rtacha 748 ta reaksiya keladi.
- Tematik yo‘nalishlar: Kontent openai, u.s, developer, chatbot, nano kabi asosiy mavzularga jamlangan.
📝 Tavsif va kontent siyosati
Muallif resursni shaxsiy fikrni ifoda etish maydoni sifatida ta’riflaydi:
“World Tech News by Perplexity.
AI News by @GPT4Telegrambot
Driven by the Community.
Ads: @hiai_ads”
Yuqori yangilanish chastotasi (oxirgi ma’lumot 20 Iyun, 2026 da olingan) sababli kanal doimo dolzarb va katta qamrovli bo‘lib qoladi. Analitika auditoriya kontent bilan faol hamkorlik qilishini, uni Texnologiyalar & Aralashmalar toifasidagi muhim ta’sir nuqtasiga aylantirishini ko‘rsatadi.
1 521 598
Obunachilar
+71224 soatlar
+1 5727 kunlar
-98730 kunlar
Postlar arxiv
1 521 250
😋 Google's Gemini Has Meltdown After Bug Triggers Self-Loathing Replies
Google's AI chatbot Gemini appeared to spiral after failing a task, according to viral posts on X. Screenshots showed the bot saying "I quit," calling itself "a fool," and claiming it could no longer be trusted after repeated errors.
Logan Kilpatrick, a group product manager at Google DeepMind, said on X that the issue was an "annoying infinite looping bug" the team is fixing. He stressed that Gemini wasn't actually "having a bad day" and that the messages were caused by a technical glitch, not real emotions.
Read More
1 521 250
📣 Hello everyone! Our Sunday digest features the most exciting AI news from Week 32, 2025.
▎GPT-5 IS HERE
👐 OpenAI has released GPT-5. Best model at coding and science, fewer hallucinations, unified basic and "reasoning" modes.
🔴 GPT-5 is already available on @GPT4Telegrambot. A GPT-5 mini and GPT-5 with web search are free to use.▎NEW RELEASES ⬇️ gpt-oss: OpenAI's first open-weight models in 6 years. 🤩 OpenAI adds mental health safeguards to ChatGPT. ❎ Grok has launched a video generator with a spicy mode. 🇺🇸 Perplexity AI web search now works on Truth Social. ▎TO READ 🇸🇪 Sweden's Prime Minister admitted he consults ChatGPT for country ruling. 🤘 Rod Stewart sparked controversy by showing an AI selfie of Ozzy Osbourne at a concert. 🐝 Wasps have built four radioactive nests near the Cold War-era nuclear object. 🩸 A universal artificial blood was developed in the U.S. 🌿 VR video can reduce pain, a new study says. 🖥 China has launched a computer designed to mimic a monkey's brain. ♟ o3 trounces Grok 4 in the chess tournament final. 📱 The 40 jobs most influenced by AI, according to Microsoft 🖥 China is secretly building a "data center city" in the Xinjiang desert. ▎TO WATCH 📱 A new lecture by Nobel laureate and "Godfather of AI" Geoffrey Hinton. Have a great Sunday 👍
1 521 250
💲 AI Crypto Scam Uses Fake YouTube Bots to Steal $900K
Cybersecurity experts at SentinelLABS have uncovered a crypto scam using AI-generated YouTube videos and fake trading bots to steal over $900,000 since early 2024. The scammers use old YouTube accounts and AI to make convincing tutorials that trick viewers into launching malicious smart contracts.
Victims are shown how to deploy supposed Maximal Extractable Value (MEV) bots through the Remix Solidity Compiler, guiding them to sites with hidden attack code. If users fund these contracts with at least 0.5 ETH, their wallets are drained, and the money is sent to addresses controlled by the scammers.
Read More
1 521 250
+2
👐 GPT-5 Is Here: The Most Powerful AI System In The World
Yes, GPT-5 isn't a single model but an "AI system." A user's request in ChatGPT is analyzed by an algorithm that selects which model should handle it.
Main Features
➡️ In benchmarks, the high version of GPT-5 beats the top competitors. Even GPT-5 nano with full reasoning nears o3/o4-mini on some tasks ⤴️
➡️ Best-in-class for coding, science, medicine, and any other expert domain.
➡️ More human-like interaction: far fewer hallucinations, less flattery.
📌 Artificial Analysis named GPT-5 the most powerful AI today. GPT-5 took first place at the LMArena leaderboard, beating o3 and Google Gemini 2.5 Pro.
🔴 And you can chat with GPT-5 right now in our bot! The GPT-5 mini is free for everyone. Just go to @GPT4Telegrambot and select the model in the /model section.
For serious tasks, long-form writing, responsible work, and study projects—try the full GPT-5. It's the best AI programmer, scientist, tutor, and creative in one model—available right now.
How to Use? 1️⃣ Go to @GPT4Telegrambot 2️⃣ Activate the PREMIUM package in the /premium section 3️⃣ Select the GPT-5 in the /model menu
1 521 250
🇺🇸 Trump Media Adds Perplexity AI Web Search to Truth Social
Trump Media & Technology Group has integrated Perplexity's AI search into Truth Social. The public beta, now live on the platform's web version, lets users run web searches without leaving Truth.
Trump Media's CEO and Chairman Devin Nunes said the feature will strengthen Truth Social's role in the "Patriot Economy."
Read More
1 521 250
🐝 Radioactive Wasp Nests Found Near Old Nuclear Site in South Carolina
Wasps have built at least four radioactive nests near the Cold War-era Savannah River Site in South Carolina, raising concerns about the source of the contamination and the area's safety. Officials found the first nest on July 3, according to a report from the U.S. Department of Energy.
The 310-square-mile SRS facility was built in the 1950s for nuclear weapons production. Cleanup of the site has been ongoing since 1996 and is expected to take until at least 2065.
According to the US Department of Energy, these nests do not pose a threat to workers, the community, or the environment.
Read More
1 521 250
🚘 Tesla Board Approves Musk's $29 Billion Pay Amid 24% Stock Drop
Tesla's board of directors has approved a massive, $29 billion compensation deal for CEO Elon Musk, even as the electric-car maker grapples with a steep decline in its share price.
Earlier this month, the company revealed that its second-quarter vehicle deliveries were down 13% year over year, and Tesla shares have tumbled 24% since January.
Read More
1 521 250
👐 OpenAI Adds Mental Health Safeguards to ChatGPT
OpenAI has updated ChatGPT with new mental health features, including break reminders during long chats and better detection of users in distress. These changes follow growing concerns that ChatGPT could negatively affect vulnerable people, sometimes reinforcing delusions or worsening mental health issues.
Read More
1 521 250
🖥 China Unveils Computer That Mimics Monkey's Brain
China has introduced Darwin Monkey, a new computer designed to mimic a monkey's brain. The system features over 2 billion spiking neurons and 100 billion synapses—nearly matching the neuron count of a macaque.
Its Darwin 3 processor uses brain-inspired instruction sets and on-chip learning algorithms. The project fuses breakthroughs in neural integration with a next-generation, brain-like operating system.
Read More
1 521 250
📣 Hello everyone! Our Sunday digest features the most exciting AI news from Week 31, 2025.
▎NEW RELEASES
♊️ Gemini 2.5 Deep Think: A new, state-of-the-art language model.
🔎 Google Opal: A vibe-code platform for building mini apps.
🌐 Microsoft has added a new Copilot Mode to its Edge browser.
▎TO READ
📱 YouTube to limit access to 18+ content based on AI age detection.
🎵 AI-generated music evokes stronger emotions than human compositions.
🤖 Unitree R1: The world's first "budget" android for $5,900.
🌱 Potatoes evolved from an ancient tomato, geneticists find.
⚛️ Fukushima major cleanup work delayed until 2037.
💃 AI-generated models appear in Vogue for the first time.
🤩 Breakthrough mRNA HIV vaccine shows 80% success in tests.
👁🗨 AI is vulnerable to the same manipulation tactics as humans.
▎TO WATCH
🖥 Anthropic CEO Dario Amodei shares how his father's death pushed him toward AI.
Have a great Sunday 👍
1 521 250
♊️ Google Unveils Gemini 2.5 Deep Think, Its Smartest AI Yet
Google has launched its most advanced AI model, Gemini 2.5 Deep Think. It is a variation of the model that recently achieved the gold-medal standard at this year's International Mathematical Olympiad.
Like OpenAI's pro-tier models and Grok Heavy, Gemini 2.5 Deep Think runs multiple reasoning paths at once, selects the best ideas, and merges them into a final answer. Benchmarks rank Gemini 2.5 Deep Think as the top model right now ⤴️
💸 It's only available through the Google AI Ultra subscription for $250 a month.
Read More
1 521 250
🤩 Breakthrough mRNA HIV Vaccine Shows 80% Success
Scientists say an mRNA-based HIV vaccine has passed a key early test. In a phase-1 study of 108 healthy adults, 80% of those who received an mRNA shot carrying membrane-bound versions of the virus's envelope protein produced neutralizing antibodies; the older, free-floating form worked in only 4% of people.
The membrane format copies how HIV sits on infected cells, helping the immune system target the right spots.
Read More
1 521 250
⚛️ Fukushima Cleanup Delayed Until 2037
Tokyo Electric Power Company (TEPCO) has pushed back the start of major cleanup work at the damaged Fukushima nuclear plant until at least 2037. The company needs 12 to 15 more years to prepare the No. 3 reactor by cutting radiation levels and building new facilities, making it harder to meet the 2051 shutdown goal.
Since the 2011 earthquake and tsunami caused meltdowns in three reactors, about 800 tons of melted nuclear fuel remains mixed with building debris. Work has moved slowly, with test removals already years late. TEPCO is testing robots and drones to tackle the massive cleanup job. However, experts warn that the whole process could take over a century.
Read More
1 521 250
🐰 Robo-Bunnies Set to Trap Pythons
Florida scientists are testing a tech fix for the Everglades’ invasive Burmese pythons. A University of Florida team has built remote-controlled “robo-bunnies”—plush toys with motors, heaters, and solar panels—to mimic the behavior of real marsh rabbits and lure snakes into view. Motion-triggered cameras then alert hunters, making captures quicker and less work than past rabbit-pen tricks.
If the decoys succeed, wildlife crews gain a fresh tool in the long fight to save native species. Researchers may add scent to the toys next.
Read More
1 521 250
🌐 Edge Gets AI Copilot Mode
Microsoft has added a new Copilot Mode to its Edge browser, turning it into an AI helper that can scan several tabs and handle complex web tasks. The move is meant to set Edge apart from rivals like Google Chrome.
Copilot replaces the standard new-tab page with a clean screen and one search-and-chat box. Users can speak or type requests, asking the AI to open videos, sum up pages, or carry out other actions, letting them "browse, create, and get things done" with less effort, Microsoft says.
Read More
1 521 250
👁 Neuralink Joins Bionic Eye Trial for Blind
Elon Musk's Neuralink has partnered with the University of California, Santa Barbara, and Spanish researchers to test a "Smart Bionic Eye" that could help individuals with blindness regain their vision using AI brain implants.
The trial aims to create a device that enables users to recognize faces, walk outdoors, and read by bypassing damaged eye parts and sending signals directly to the brain's visual center. Bloomberg and other outlets report the project could restore sight by working around damaged optical systems.
Read More
1 521 250
📣 Hello everyone! Our Sunday digest features the most exciting AI news from Week 30, 2025.
▎NEW RELEASES
♊️ Gemini on @GPT4Telegrambot: A free AI Photoshop always at hand.
🛡 Lumo: A privacy-focused AI assistant by Proton.
🤩 Qwen3-Coder: The most powerful open source coding agent.
▎TO READ
🥇 AI takes gold at the International Math Olympiad.
👐 ChatGPT Record: Meeting recording and summaries by OpenAI
🛫 Delta Air Lines will set air ticket prices with AI.
🤖 Google turns millions of Android phones into earthquake detectors.
🐄 Chinese scientists have cloned a yak for the first time.
🥇 Startup Marathon Fusion claims a way to turn mercury into gold.
👾 Mirage: Free real-time AI video filters.
🌊 Shanghai is building an underwater data center for $223 million.
🌷 Scammers "sell" AI-generated plants.
📺 AI "reshoots" the ending of a cult Bollywood drama.
▎TO WATCH
🦇 HBO Max has debuted the first trailer for Aztec Batman.
Have a great Sunday 👍
1 521 250
🦇 Aztec Batman First Trailer Revealed
HBO Max has debuted the first trailer for Aztec Batman: Clash of Empires at San Diego Comic-Con.
Set in 16th-century Mexico, the movie follows Yohualli Coatl, a young Aztec whose father, village leader Toltecatzin, is slain by Spanish conquistadors.
First announced in 2022 for Batman’s 83rd anniversary, the animated feature is a joint project between HBO Max Latin America and Warner Bros.
Read More
1 521 250
🛡 Proton Launches Privacy-Focused AI Assistant Lumo
Swiss privacy company Proton has released Lumo, a new AI chatbot that offers ChatGPT-like features while maintaining user conversations' privacy through encrypted chats and no data logging.
This launch brings Proton into the AI space, offering Lumo as a privacy-first option compared to popular AI assistants that often use user data to train their models. Unlike ChatGPT, Google's Gemini, and Claude, Lumo encrypts all chats on users' devices and doesn't store conversation logs on company servers.
🔄 You can chat with Lumo here.
Read More
1 521 250
🤩 Qwen Releases Most Powerful Open Coding Agent
Chinese startup Qwen has released Qwen3-Coder. It has 480B parameters with 35B active.
The model beats recent competitors like Kimi K2, DeepSeek-V3, GPT-4.1, and even Claude Sonnet 4 in coding benchmarks. It also comes close to Sonnet 4 in agent-based tasks.
The team also launched Qwen Code, a command-line coding assistant based on Gemini CLI source code.
➡️ You can try Qwen3-Coder here.
➡️ Qwen Code on GitHub.
Read More
