You can now edit images with Grok 3

Welcome back! Hope you had a very nice weekend! Here’s what you need to know about AI today:

👉 xAI added a new image editing feature to Grok 3

👉 Real-time AI video features are coming to Gemini

👉 Perplexity AI wants to buy TikTok!

and many more!

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 5.1 minutes

What’s going on: xAI has introduced a new feature allowing users to edit images using its AI chatbot, Grok 3. This update enables customization of specific image elements through simple text prompts, eliminating the need for advanced editing software. This is probably xAI’s answer to the experimental native image generation feature that Google added to Gemini 2.0 Flash, two week ago.

What does it mean: Now users can upload an image and instruct Grok to alter a particular section, such as changing an object or adjusting a detail and the AI will modify only that part while preserving the rest of the image. Google (with Gemini) and xAI (with Grok) are now officially the only two companies that are offering this functionality for free.

More details: 

• The next step could be full video generation, which xAI is also working on, after recently acquiring Hotshot.

• This feature is now available to both free and premium subscribers on X and Grok apps/websites.

• The same feature has also been available for a while in dedicated visual generation tools like OpenAI’s Dall-E 3 (for some users) and Midjourney (for its special users and through its Full Editor).

• How to use Grok’s image editing feature: Go to Grok or X’s Grok section → click on “Edit Image“ button in prompt box → upload your image → ask it to modify the image as you wish (you can add/subtract objects, change the image’s style, change lighting, change colors, and so on).

What’s going on: Google has started rolling out new features for Gemini Live, its conversational AI assistant. These updates, part of the Project Astra initiative, allow Gemini to analyze smartphone screens and camera feeds in real time. The rollout targets Gemini Advanced subscribers under the Google One AI Premium plan, with initial access reported on Android devices.

What does it mean: These features enable users to ask questions about what’s on their screen or in their camera’s view, with Gemini providing contextual responses based on live data. The screen-sharing capability builds on an earlier static screenshot feature, now offering continuous, real-time access to on-screen content.

More details: 

• Users can now discuss a webpage or app interface as they navigate it. The live video feature lets Gemini interpret camera input, such as helping a user choose a paint color for pottery by analyzing the object in view.

• Google first demoed these capabilities nearly a year ago, and their release follows a promise made at MWC 2025 to deliver them by the end of March.

• If this feature is not yet available to you, be patient. Google has started to gradually unlock this feature for more users and it is only being tested by some users like the one who shared the experience in a Reddit post. As of today, there has been no word on expansion to free-tier users or other platforms like iOS.

🔍 Perplexity AI has expressed interest in acquiring TikTok in a recent blog post, proposing to rebuild its algorithm with transparency and integrate its AI-powered search capabilities to enhance the platform.

🚀 Oracle has launched the AI Agent Studio as part of its Fusion Applications Suite, enabling businesses to create, customize, deploy, and manage AI agents with advanced tools, seamless integrations, and built-in security at no additional cost.

🐜 Ant Group, Alibaba-affiliate, has developed cost-efficient AI training methods using a mix of Chinese and US semiconductors, including chips from Alibaba, Huawei, and AMD, cutting costs by 20% while reducing reliance on Nvidia.

⌚ Apple is reportedly exploring adding cameras to the Apple Watch (both standard Series and Ultra models) by 2027 to support AI features and could be used for in-air hand gesture detection and working with devices like the Apple Vision Pro to enhance spatial audio or detect environmental changes.

✋ FuriosaAI, South Korean AI chip startup, has turned down an $800 million acquisition offer from Meta, choosing to remain independent and continue developing its AI chips, Warboy and RNGD.

💰 Browser Use, a startup that makes websites more readable for AI agents by converting them into a digestible and text-like format, has raised $17 million in seed funding.

🦁 Anthropic appears to be using Brave Search to power the web search feature recently rolled out for its Claude chatbot. Evidence for this includes Anthropic adding "Brave Search" to its subprocessor list and identical citations returned by Claude and Brave for at least one search.

🤝 Microsoft is exploring a "training-time provenance" research project to credit and potentially compensate contributors to AI training data, aiming to estimate the influence of specific data (like photos and books) on AI model outputs.

💬 Meta is testing an AI-powered feature on Instagram that suggests AI-generated comments for users to post on their friends' photos.

🤖 1X, a Norwegian robotics startup, plans to begin testing its humanoid robot, Neo Gamma, in a few hundred to a few thousand homes by the end of 2025, utilizing teleoperators to assist with the robot's movements and collect data for AI model training.

AI + Learning guide

ByteDance - Senior Data Scientist (Multiple Positions)

Outlier - Coding Expertise for AI Training

Webflow - Staff Product Manager, AI Code Gen

LinkedIn - Data Architect - AI & Automation

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!