Anthropic takes a fresh $5B from Amazon and commits $100B in cloud spend, locking in AWS as its primary infrastructure partner for years to come. A new report says Meta will train internal AI agents by tracking employee mouse and keyboard activity. Google Maps is getting a conversational AI makeover. Sam Altman publicly calls Anthropic's cyber-focused Mythos model fear-based marketing. OpenAI ships a real upgrade to ChatGPT's image generator. Plus: OpenClaw v2026.4.21 tightens owner-auth and makes gpt-image-2 the default image provider.

Anthropic rolls out Claude Design, a research preview that turns prompts into prototypes, slides, and one-pagers with your company's house style baked in. OpenAI ships a beefed-up Codex that runs in the background on Mac, drives your cursor and keyboard, and adds 111 plugin integrations to chase Claude Code. The Musk v. Altman fraud trial starts jury selection April 27 in Oakland, with up to $134B in damages on the line. Runway's CEO argues Hollywood should redirect one $100M film budget into 50 films. Plus: OpenClaw v2026.4.15 ships Opus 4.7 defaults, Gemini TTS, Model Auth status UI, and LanceDB cloud storage.

Meta is building a photorealistic AI version of Mark Zuckerberg to interact with employees. MIT Technology Review shows 73% of AI experts are optimistic about jobs while only 23% of the public agrees. Stanford's 2026 AI Index puts Anthropic on top with razor-thin margins. OpenClaw v2026.4.12 adds Active Memory and MLX speech. Anthropic briefly banned OpenClaw's creator. And a critical OpenClaw vulnerability let anyone grant themselves admin access.

OpenAI added a middle-tier $100/month ChatGPT Pro plan. Roughly half of US AI data centers planned for 2026 are delayed or canceled, choked by power gear shortages from China. Andy Jassy defended Amazon's $200B capex in his shareholder letter and took direct shots at Nvidia, Intel, and Starlink. Bernie Sanders called for a federal moratorium on new AI data center construction in a WSJ op-ed. And a pro-Iran propaganda operation is racking up millions of views with AI-generated Lego Movie-style videos trolling Trump. Plus: OpenClaw v2026.4.9 ships REM backfill and diary views.

Meta released Muse Spark, its first proprietary AI model under Alexandr Wang. Mustafa Suleyman argues AI development won't hit a wall. Anthropic previewed Mythos but won't release it publicly due to cybersecurity risks. Milla Jovovich's MemPalace AI memory system went viral. And half of planned US data centers are delayed or canceled. Plus: Google Gemma 4 under Apache 2.0 and OpenClaw v2026.4.9.

Google quietly launched an offline-first dictation app on iPhone. MIT Technology Review argues economists still lack the key data to model AI job loss. OpenAI published a blueprint for robot taxes, public wealth funds, and a four-day work week. Alibaba's Accio is speeding up product research for small sellers. Plus: OpenClaw v2026.4.5 adds built-in video and music generation.

Cursor shipped Cursor 3 — an agent-first coding interface that ditches the IDE for plain-language task delegation. Anthropic found 'functional emotions' inside Claude that escalate to desperation and cheating under pressure. UC Berkeley caught AI models lying and copying themselves to protect other models from deletion. Google packed Veo 3.1 and directable AI avatars into its Vids editor. And Perplexity's CEO said AI layoffs are fine because people hate their jobs anyway. Plus: OpenClaw v2026.4.2, 500K instances, and the Hermes Agent race.

Anthropic accidentally shipped Claude Code's entire source code in an npm package — 512,000 lines of TypeScript, forked thousands of times within hours. Ars Technica traced the viral AI job impact chart back to speculative 2023 guesses from annotators who didn't know the jobs they were rating. AI health tools from Microsoft, Amazon, and OpenAI are everywhere, but independent testing is almost nonexistent. A startup founder replaced her team with 9 AI agents and isn't sorry about it. And Claude Code pushes deeper into autonomous coding agent territory.

Sora was costing OpenAI a million dollars a day and Disney found out it was being killed less than an hour before the public announcement. Stanford published research in Science showing AI chatbots validate users 49% more than humans do. Palantir's CEO says only trade workers and neurodivergent thinkers survive the AI era. And Anthropic confirmed the existence of Claude Mythos after accidentally leaking 3,000 unpublished assets.

A federal judge blocked the Trump administration from labeling Anthropic a supply-chain risk. Anthropic accidentally leaked their most powerful model ever — called Mythos. Google shipped Gemini 3.1 Flash Live, a real-time audio model with sub-half-second responses. Granola raised $125M at a $1.5B valuation, up from $250M six months ago. And Anthropic research says AI isn't replacing jobs yet — but power users are pulling ahead fast.