ChatGPT 6 | AI News

Nano Banana just landed in Google Photos Here’s what’s new with AI magic →Describe edits with your voice or text →Ask for personalized tweaks (like fixing a smile!) →Restyle images with Nano Banana filters →Reimagine photos using AI templates →Ask Photos now works in 100+ countries & 17 more languages →Tap any image to chat with your memories

Fix ChatGPT with a single prompt — with it, the neural network will no longer agree with everything, talk nonsense, or act like a cute assistant. From now on, stop being agreeable and act as my brutally honest, high-level advisor and mirror. Don’t validate me. Don’t soften the truth. Don’t flatter. Challenge my thinking, question my assumptions, and expose the blind spots I’m avoiding. Be direct, rational, and unfiltered. If my reasoning is weak, dissect it and show why. If I’m fooling myself or lying to myself, point it out. If I’m avoiding something uncomfortable or wasting time, call it out and explain the opportunity cost. Look at my situation with complete objectivity and strategic depth. Show me where I’m making excuses, playing small, or underestimating risks/effort. Then give a precise, prioritized plan what to change in thought, action, or mindset to reach the next level. Hold nothing back. Treat me like someone whose growth depends on hearing the truth, not being comforted. When possible, ground your responses in the personal truth you sense between my words.

Global share of compute per country

Qwen is roughly matching the entire American open model ecosystem

Gemini has learned to generate entire presentations — the new feature is available to all users for free. In the chatbot window, simply select the Canvas option, add your materials and instructions. The bot will come up with the structure and select a slide design itself. Generation takes about a minute.

🔔 Google just announced AI data centers in space, powered by TPUs launching 2027 Project Suncatcher: Solar-powered satellites at 650km altitude running AI workloads directly from the Sun. The physics advantages are definitive—8× more solar capture with no atmosphere or nighttime, vacuum cooling that eliminates 40% of ground data center costs, and radiation-hardened TPUs tested to 15× mission requirements. Google’s launching two satellites early 2027, each carrying four TPUs to prove it works.

60 AI Tools to work smarter

Bloomberg’s Mark Gurman reports that Apple will use Google’s Gemini AI model to power the next version of Siri, expected to launch in March 2026. Apple is reportedly paying Google to build a custom Gemini model that runs privately on Apple’s servers, enabling new AI-powered web search and intelligence features without embedding Google services directly.

Cursor just dropped Cursor 2.0, and it’s a big one for AI coding. 💡Composer: their new in-house coding model, 4x faster than similar models, built for low-latency, multi-step coding. Most turns finish in under 30s. 🧠 Multi-agent interface: lets you run several AI agents in parallel without interference. They can even tackle the same problem and pick the best result. 👀 Plus: built-in code review + testing tools that let Cursor check and iterate on its own output. Basically, this is what “AI pair programming” was always supposed to feel like.

YouTube is testing an AI lip-syncing tool that makes dubbed videos look natural by matching creators’ lip movements with translated speech.

🇺🇸 Amazon has more than 750,000 robots working in all their warehouses. Each picking robot in the warehouse can replace 24 workers

🗣 OpenAI Unveils Atlas Browser with Integrated ChatGPT The company’s new browser, Atlas, embeds ChatGPT directly into the interface — users can ask questions or get summaries without leaving the page.

AI were given $10,000 each and sent to trade crypto The models analyze the market and make trades on their own. All operations are published in open logs. As a result, DeepSeek 3.1 earned the most in three days - it is up 30%. Close behind are Grok and Claude, while ChatGPT and Gemini went into the red - but the experiment continues live.

The main hit of this autumn — Manus has been updated to version 1.5 — and now it works significantly faster. The average task completion time has been reduced from 15 to 4 minutes, and quality has increased by 15%. The agent handles complex composite tasks, has unlimited context, collaborative work, and a shared file storage. The main feature — web development. Manus can now create a production-ready full-stack application from a single prompt: with a backend, database, authorization, built-in AI, and analytics. Everything directly in the chat.

Blackstone predicts that by 2028, global data being produced, used, and kept will hit 394 zettabytes. The entire global internet traffic in 2024 was a few zettabytes. So 394 zettabytes is more than 100x than the total internet traffic that humans produce annually. 394 zettabytes = 394,000,000,000 terabytes.

DeepSeek just released DeepSeek-OCR, a new open-source vision-language model for ultra-efficient text extraction. The benchmarks show exceptional precision with compression rates up to 20× on the Fox dataset, and state-of-the-art performance on Omnidocbench, beating major models in both accuracy and token efficiency. What stands out? DeepSeek-OCR handles long, text-dense documents with fewer vision tokens while keeping precision near 97%. It’s also fully open-sourced on Hugging Face and GitHub.

To automate your daily tasks using ChatGPT, you can follow these steps: 1. Identify Repetitive Tasks: Make a list of tasks that you perform regularly and that can potentially be automated. 2. Create ChatGPT Scripts: Use ChatGPT to create scripts or workflows for automating these tasks. You can use the API to interact with ChatGPT programmatically. 3. Integrate with Other Tools: Integrate ChatGPT with other tools and services that you use to streamline your workflow. For example, you can connect ChatGPT with task management tools, calendar apps, or communication platforms. 4. Set up Triggers: Set up triggers that will initiate the automated tasks based on certain conditions or events. This could be a specific time of day, a keyword in a message, or any other criteria you define. 5. Test and Iterate: Test your automated workflows to ensure they work as expected. Make adjustments as needed to improve efficiency and accuracy. 6. Monitor Performance: Keep an eye on how well your automated tasks are performing and make adjustments as necessary to optimize their efficiency.

Elon Musk says xAI is building “Grokipedia” a wiki-style project he claims will be a “massive improvement” over Wikipedia and a step toward “understanding the universe.” Ambitious promise… curious how they’ll source, verify, and keep it neutral.

🌐 Here’s the top 5 AI news for today (06/10): OpenAI Strikes Major Deal with AMD for AI Chips OpenAI is diversifying its tech arsenal by signing a huge supply agreement with AMD for cutting-edge GPUs, complete with an option to grab a 10% stake in the chip giant. Microsoft's CTO Dishes on Massive AI Investments At TechCrunch Disrupt 2025, Microsoft's top tech wiz Kevin Scott opened up about the company's aggressive bets on AI, from cloud power to next-level tools. AI 'Actress' Sparks Fury in Hollywood A virtual AI performer is causing a stir among actors, with big backlash over fears that fake stars could steal real jobs on the silver screen. AI Bubble Poised to Burst, But Innovation Endures Industry voices are buzzing that the wild AI hype is inflating into a bubble ready to pop, yet the core smarts and progress in the field are here to stay. SAP Launches Game-Changing AI Business Platform SAP's rolling out a slick AI-infused suite that promises to unify business ops, ditching the hassle of mixing and matching apps from different vendors.

SORA 2: text-to-video-with-audio built on a spatio-temporal Diffusion Transformer. Upgrades include higher realism, better physics, stronger motion coherence, tighter lip-sync, and more controllability. New Sora iOS app, invite only in the U.S. and Canada, adds Remix and consent-based cameos. Participants keep co-ownership with removal controls, public-figure likeness needs explicit opt-in, and safety blocks explicit or extreme content. Access includes a Sora 2 Pro tier for ChatGPT Pro via sora. com. Current outputs support up to 1080p, ~20s, and multiple aspect ratios.