AI TodayGet the app

Image Generation

Curated AI news · 24 stories

Image Generation

Google Images gets a Pinterest-like redesign focused on discovery

Google Images has been redesigned to resemble Pinterest by offering a dynamic, browsable gallery tailored to users' interests, allowing them to save images into collections for inspiration. Additionally, Google is introducing an AI image generation feature within Search, enabling users to create custom visuals from text prompts. These updates aim to enhance user engagement and keep users within Google's ecosystem for image discovery and creation.

Image Generation

The Google Images homepage will recommend photos even before you search

Google is updating the Google Images homepage to display a dynamic, personalized gallery of images before users initiate a search, similar to platforms like Pinterest. This feature will roll out soon for signed-in desktop users in the US. Additionally, Google Search will introduce AI-generated images using its Nano Banana 2 Lite model to help users visualize concepts, with details on prompt triggers and content controls still pending.

AI Research

Meta’s AI detector can’t catch its own cropped fakes

Meta introduced an AI image detector designed to identify images generated by its Muse Image model, using an invisible watermark called Content Seal. However, Reuters found that cropping these images caused the detector to miss over half of them, revealing a limitation in the watermark's robustness against common edits. Meta acknowledged the detector is still in preview and that heavy cropping can remove the watermark signal, highlighting challenges faced industry-wide in reliably detecting AI-s

AI Research

Meta’s AI Detector Can’t Detect Images It Generated Itself, Report Finds

Meta introduced an AI image generation model called Muse Image along with a watermarking system named Content Seal, designed to identify AI-generated images even after modifications like cropping. However, a Reuters report found that Meta's AI detection tool failed to recognize over half of the cropped images as AI-generated, highlighting ongoing challenges in AI content detection. This shortfall underscores the broader difficulty in keeping pace with the rapid growth of AI-generated deepfakes,,

Image Generation

The Public Got So Mad at Meta’s New AI Photo Tool That It’s Scrapped Already

Meta launched an AI photo generation feature that automatically used face data from public Instagram accounts without explicit consent, leading to significant public backlash. Within just over three days, Meta removed the feature following criticism from users and organizations like SAG-AFTRA, which emphasized the need for clear opt-in consent for such uses. This incident reflects ongoing challenges in AI product releases related to privacy and consent.

AI Research

Google’s SynthID watermark just debunked its first high-profile deepfake: a hoax image of Mitch McConnell

Google's SynthID watermarking technology successfully identified a viral AI-generated image falsely depicting Senator Mitch McConnell in a hospital, marking its first significant real-world application. The invisible watermark embedded in the image data remained intact despite screenshots and cross-platform sharing, enabling verification by fact-checkers like Snopes. While SynthID is effective for images generated by participating tools, its utility is limited by voluntary adoption and does not涵

AI Research

If You Have a Public Instagram Account, You Might Be Surprised at What AI Users Can Now Do With Your Face

Meta has introduced an AI image generator called Muse Image that integrates with Instagram, allowing users to generate images based on photos from Instagram accounts. The tool can create images using data from both the user's own account and from other public Instagram profiles, raising privacy concerns about the use of individuals' likenesses without their consent. Similar AI tools have been released by other companies but have faced issues related to misuse and content moderation.

Image Generation

Meta just launched a new AI generator, Muse Image, and users are already pushing back over use of their photos

Meta has introduced Muse Image, a new AI-powered image generator accessible via the Meta AI app, Instagram Stories, and WhatsApp. The tool offers features like preset prompts, image editing, and integration with Facebook Marketplace for applications such as creating ads and visualizing furniture placement. While the service is free for basic use, advanced usage requires a subscription, and Meta is also developing a related AI video generator called Muse Video.

Image Generation

Meta’s new Muse Image model can pull other Instagram users into AI photos

Meta has introduced Muse Image, an AI model developed by its Superintelligence Labs, which powers image generation features across Meta's platforms including Instagram and WhatsApp. The model allows users to include other Instagram users in AI-generated images by tagging their accounts, using public photos to create visuals, with user controls for content reuse. Additional capabilities include transforming images with prompts, redesigning rooms from online images, and editing photos directly, as

AI Research

Meta debuts Muse Image, its first AI image model built under Alexandr Wang’s lab

Meta has introduced Muse Image, its first AI image-generation model developed under Alexandr Wang's Superintelligence Labs. The model is integrated into Meta AI chatbot, Instagram, and WhatsApp, allowing users to create or modify images, including generating pictures of friends from public posts with opt-out options. Advertisers will gain access soon, and Meta plans to offer AI model access to external developers via cloud services while continuing to invest heavily in AI infrastructure.

LLM & Text Generation

Build a serverless image editing agent with Amazon Bedrock AgentCore harness

Amazon Bedrock's AgentCore harness enables building a serverless image editing AI agent that processes natural language commands to modify photos. The agent orchestrates multiple AI models and tools within isolated microVMs, managing memory and tool routing without custom orchestration code. The solution includes features like model switching, persona overrides, conversation memory, and is deployable via AWS CDK with a React frontend.

Machine Learning

PRX Part 4: Our Data Strategy

The article details the data strategy behind PRX, focusing on assembling a large and diverse dataset for pre-training the model. It emphasizes using a mix of public and internal datasets, re-captioning images with a vision-language model, and prioritizing breadth and diversity over individual image quality to teach the model about visual concepts. The approach balances leveraging existing curated sources with the goal of comprehensive coverage to support effective pre-training.

AI Research

Midjourney turns the tables, demanding the studios suing it reveal their AI

Midjourney, an AI image generator sued by major studios for allegedly infringing on copyrighted characters, has requested a US court to compel these studios to disclose their own AI training practices and data. The company argues that if the studios use similar AI training methods, it would support Midjourney's defense of fair use and the legal doctrine of "unclean hands." The studios reject this request, calling it a fishing expedition, while the case could have significant implications for AI,

AI Research

Midjourney wants Hollywood studios to reveal the details of their AI usage

Midjourney, an AI startup, is involved in legal disputes with Disney, Universal, and Warner Bros. over alleged copyright infringement related to its AI-generated images of copyrighted characters. Midjourney is seeking to compel the studios to disclose their own use of AI, arguing that such information is necessary for its defense and that the studios may be engaging in similar practices internally. The studios contend they want to prevent unauthorized copying of their content but do not intend a

Politics

Are Those Fake Books on the New Air Force One AI-Generated?

A photo posted by White House Press Secretary Karoline Leavitt aboard the newly refurbished Air Force One revealed shelves with books having generic titles like "Library," "Arts," and "Architecture," leading to speculation about whether the image was AI-generated. Tests using Google's Gemini AI chatbot found no SynthID watermark indicating Google's AI tools were used, but the possibility of other AI tools remains; however, it is more likely the books are physical but fake with generic covers.