OpenAI GPT-Image-2 Leaks with Text That Works
Three image generation models appeared on LM Arena under codenames maskingtape-alpha, gaffertape-alpha, and packingtape-alpha. The AI community identified them as OpenAI's unreleased GPT-Image-2.
Blog
Deep dives, tutorials, and analysis for AI-powered creators.
Sync-3 Ships 4K AI Lipsync With Full-Shot Processing
Sync, the Y Combinator-backed lipsync startup, launched Sync-3 on April 6. The new model processes entire shots at once, producing 4K native output with built-in obstruction detection.
App Store Sees 84% Surge as AI Coding Tools Take Off
Apple App Store reversed a years-long decline in new app submissions, with AI-powered vibe coding platforms like Claude Code and OpenAI Codex enabling non-programmers to build apps.
Google Eloquent Brings Offline AI Dictation to iPhone
Google quietly launches Eloquent, a free Gemma-powered dictation app for iOS with offline transcription, filler word removal, and text transformation tools.
Meta Plans Open-Source Mango Multimedia AI Generator
Meta confirmed it will release open-source versions of its upcoming Mango multimedia generator and Avocado LLM, extending its open-weight strategy to next-gen frontier models.
GitHub Copilot Rubber Duck Pairs AI Models for Code Review
GitHub shipped a feature called Rubber Duck for Copilot CLI that pairs two AI models from different families to review each other's code before execution.
OpenAI CFO Sarah Friar Flags Risks in Sam Altman's 2026 IPO Push
OpenAI CFO Sarah Friar has raised serious concerns about CEO Sam Altman's push for a 2026 IPO and the company's proposed $600 billion infrastructure spending plan.
AI Singer Eddie Dalton Claims 11 iTunes Chart Spots
An AI-generated singer named Eddie Dalton now holds 11 positions on the iTunes top 100 singles chart and sits at number 3 on the albums chart, exposing gaps in how platforms handle synthetic music.
AI Virtual Try-On Startups Target Fashion's $600B Return Problem
A new wave of AI startups is attacking fashion e-commerce's most expensive problem: product returns exceeding 50% in some categories.
Anthropic Blocks OpenClaw From Claude Subscriptions
Anthropic has cut off Claude Pro and Max subscription access for OpenClaw and all third-party AI agent frameworks, forcing more than 135,000 active instances onto pay-as-you-go API pricing that can cost up to 50 times more.
xAI Grok Imagine Adds Quality and Speed Modes
xAI has rolled out two new generation modes for Grok Imagine on the web, giving users direct control over the tradeoff between image detail and generation speed. A third Pro mode with 1080p output is confirmed for later this month.
Netflix Open-Sources VOID for Physics-Aware Video Removal
Netflix has released VOID (Video Object and Interaction Deletion), its first public AI model. The open-source tool removes objects from video while preserving physically plausible interactions.
AI Voice Clones Target Folk Musician on Spotify and YouTube
Murphy Campbell found AI-generated songs on her Spotify profile she never recorded, then faced copyright claims from cloned recordings on YouTube.
Claude Computer Use Arrives on Windows for Task Automation
Anthropic expanded Claude computer use to Windows on April 3, letting the AI open apps, navigate browsers, and work on local files through Claude Cowork and Claude Code.
DeepSeek V4 Will Run Exclusively on Huawei Chips
DeepSeek V4 will run exclusively on Huawei Ascend 950PR processors, marking a milestone in China's push to build advanced AI without American chip technology.