Turn Trash Photos into Treasure: Google's Pomelli Now Generates Studio-Grade Product Shots for Free

The new Photoshoot feature in Google Labs' free marketing tool uses business context and AI to transform basic product pictures into professional lifestyle and studio imagery.

TinyFish Web Agents Outperform OpenAI and Anthropic on Hard Web Tasks, Company Releases All 300 Benchmark Runs Publicly

The platform scored 90% on Mind2Web—29 points ahead of Operator—and launched an open-source cookbook with ready-to-use automation recipes for everything from manga tracking to logistics intelligence.

Lambda and Oumi Partner to Let Enterprises Build Custom AI Models 100x Faster

The complete stack combines Oumi's automated model development platform with Lambda's GPU infrastructure, promising 70% cost savings and full data control.

No Screen, No Problem: Inside Google's Jetpack Compose Glimmer for Transparent AI Glasses

The new design system rethinks color, typography, and motion from the ground up for displays that add light to the world rather than blocking it.

From Claude Code to Canvas: Figma Now Lets You Turn Production Code Into Editable Designs

A new capture-to-Figma capability bridges code-first exploration and visual collaboration, letting developers and designers bring live UIs directly into shared workspaces.

Your AI Agent Just Moved Into Your Messaging Apps: Manus Lands on Telegram

The company launches Manus Agents, bringing full task execution—research, file handling, voice, and multi-step reasoning—directly into chat with zero setup.

OpenAI Rolls Out Lockdown Mode for ChatGPT to Thwart Prompt Injection Attacks

New optional setting restricts how the AI interacts with external systems, while “Elevated Risk” labels give users clearer visibility into features that may introduce additional security concerns.

Seed2.0 Launches With Three Agent Models Built for Large-Scale Production Deployment

Seed2.0 has officially released its latest model series, offering three general-purpose agent models—Pro, Lite, and Mini—designed for complex real-world tasks at production scale.

Uber Just Declared War on Human Drivers—Expands Robotaxis to 4 New Countries

The ride-hailing giant adds Hong Kong, Madrid, Houston, and Zurich to its driverless roster as it pushes toward 10+ markets by year-end 2026.

Your Laundry's Worst Nightmare: A Robot That Folds Clothes Just Dropped

The company's first commercial product is available now to Bay Area residents, with two purchase options and a promise of frequent updates.

Anthropic: Claude Opus 4.6 Sabotage Risk “Very Low but Not Negligible”

Anthropic: Claude Opus 4.6 Sabotage Risk “Very Low but Not Negligible”

GLM-5 Launches With Focus on Long-Horizon Agentic Tasks and Systems Engineering

The open-source model scales to 744B parameters, introduces new RL infrastructure, and adds native document generation capabilities.

ElevenLabs Adds Emotional Intelligence to Voice Agents With Expressive Mode Launch

The update pairs a new conversational TTS model with smarter turn-taking, allowing agents to detect stress, adjust tone, and respond in more than 70 languages.

Qwen-Image-2.0 Launches: One Model, Two Tracks, Zero Compromise

The update merges generation and editing into a single system, with notable gains in text rendering, resolution, and instruction capacity.

Kling AI Launches 3.0 Model, Promising Director-Level Control for Digital Creators

The latest update focuses on consistency, longer video clips, and enhanced audio for a more streamlined creative process.

AssemblyAI Launches Universal-3 Pro: A Promptable Speech Model That Adapts to Your Instructions

AssemblyAI has introduced Universal-3 Pro, a new speech language model that it describes as the first production-quality, promptable model of its kind.

Moya Humanoid Robot Debuts with 92% Human-Like Walking Accuracy, Straddling the Uncanny Valley

A new humanoid robot named Moya has been introduced by the Chinese robotics company DroidUp

Hugging Face Shifts AI Benchmarking to a Community-Driven, Transparent Model

Hugging Face has launched a new, decentralized system for tracking benchmark results. T

Your IDE Just Got a Co-Pilot That Can Take the Wheel: Xcode Integrates Claude and Codex Agents

Apple has dramatically expanded the role of AI in its core development environment.

Community Notes Introduces \"Collaborative Notes\": An AI-Powered, Community-Edited Fact-Checking System

X has launched an experimental new feature within its Community Notes program called Collaborative Notes

SpaceX Acquires xAI, Unveils Vision for Orbital AI Data Centers and a Sentient Sun

SpaceX has announced the acquisition of xAI, forming what it describes as \"the most ambitious, vertically-integrated innovation engine on (and off) Earth.\"

The Bot Utopia Experiment: One Guy's Weekend Project Becomes a Social Network for 100,000 AI Agents

Moltbook—envisioned as a social network built by and for AI agents—has reportedly grown to over 100,000 autonomous members in a matter of days.

No More Explaining Your Job to the AI. Just Install Its Brain Implant

Cowork—the platform that brings Claude Code to everyone—introduces support for plugins.

Open-Source Speech Recognition Just Leveled Up: Qwen3-ASR Delivers SOTA Accuracy and Blazing Speed for 52 Languages

This new suite, open-sourced under the Apache 2.0 license, introduces two powerful, all-in-one speech recognition models and a novel forced alignment model, setting new benchmarks for performance, speed, and versatility.

The AI Productivity Paradox: Code Faster, But Understand Less? New Study Reveals a Learning Trade-Off