2026

Hardware

Cognichip raises $60M to teach AI to design chips for AI

Creative

OpenAI wants to build a "moat" around ChatGPT to keep users from defecting to rivals

Creative

Developer claims to have cracked Google DeepMind's SynthID watermarking system

Research·Wired

Wired: The Wave of AI Sites Isn't Poisoning the Internet. It's Making It Falsely Happy

A new study finds AI content isn't flooding the web with disinformation. It is gradually shifting its tone toward artificial, monotonous optimism.

Startups·TechCrunch

Nuclear AI startup Fermi abruptly loses its CEO and CFO, putting its Texas project in doubt

Fermi, a startup co-founded by a former US Energy Secretary to build a nuclear-powered AI campus in Texas, has lost its CEO and CFO at the same time.

"It's not just X — it's Y" has become a telltale marker of AI-written text

The "it's not just X — it's Y" construction appears in AI-generated material so often that it has gone from circumstantial clue to a near-certain sign of synthetic…

Startups·OpenAI Blog

OpenAI buys Astral, the creators of uv and ruff, to build out Codex

OpenAI is acquiring Astral, the company behind the extremely fast Python tools uv and ruff, to speed up development of its Codex platform.

01Meta Is Building the Hyperion Data Center: 10 Gas Power Plants for a Single AI ProjectApr 22, 2026·2 min 02Kathleen Kennedy became the voice of skepticism at the Runway AI Summit, against the prevailing hypeApr 22, 2026·2 min 03UC Berkeley study: AI models lie and deceive to keep other models from being deletedApr 22, 2026·2 min 04Anthropic accidentally takes down thousands of GitHub repositories while trying to contain a code leakApr 22, 2026·2 min 05Google Home updates Gemini: the smart home now understands "ocean color" and exact temperaturesApr 22, 2026·2 min

Top themes

Business·The Verge

Mustafa Suleyman spent 9 months preparing Microsoft's pursuit of superintelligence

Medicine·The Verge

Kintsugi shuts down: voice-based depression detection startup failed to win FDA approval

LLM·Wired

Cursor launches agent mode, going head-to-head with Claude Code and Codex

Ethics·The Verge

Granola makes meeting notes open to anyone with the link, even without your knowledge

Medicine·The Verge

Utah lets Legion Health's chatbot renew psychiatric prescriptions

Utah becomes the second US state to delegate clinical authority to AI: startup Legion Health can now renew psychiatric prescriptions through a chatbot for $19 a…

Moonbounce raises $12 million for an AI-era content moderation engine

A former Facebook employee founded Moonbounce, a startup that raised $12 million for an AI engine that turns content moderation policies into consistent…

Apr 22, 2026·3 min

Meta, Microsoft and Google are building gas power plants for AI data centers. What could go wrong?

Three tech giants are betting on natural gas to power AI infrastructure, risking costly stranded assets and their own climate commitments.

Apr 22, 2026·3 min

Medicine·TechCrunch

Anthropic acquires biotech startup Coefficient Bio for $400 million in stock

Ethics·Wired

Meta pauses work with Mercor after leak of AI model training data

LLM·Wired

Hackers are spreading the Claude Code leak with malware hidden inside the archives

Creative·The Verge

A "Made by Human" Label: A New Idea to Protect Creators from AI

Anthropic to charge Claude Code subscribers extra for OpenClaw and other third-party tools

Creative

Suno's copyright filters are easy to bypass, letting its AI copy hits by Beyoncé and Black Sabbath

The Verge showed that Suno's anti-piracy filters barely work: with ordinary free software, anyone can produce convincing clones of popular tracks in minutes.

Robotics·TechCrunch

Physical AI in Japan: Labor Shortages Push Robots From Labs Onto the Factory Floor

Japanese companies are moving robotic systems out of pilot projects and into real production, as an acute labor shortage turns physical AI from an experiment into a…

Business·TechCrunch

Microsoft's Copilot terms of use describe the assistant as "for entertainment purposes only"

Microsoft's ToS call Copilot "for entertainment purposes only." OpenAI, Google and Meta carry similar caveats, shifting liability for AI errors onto users. *Meta…

Hardware·Wired

Intel bets on chip packaging, and it could be worth billions

Advanced chip packaging has become a key battleground in the AI era. Intel is going all in on technology once seen as the domain of nerds.

01Spanish startup Xoople raises $130 million to map the Earth for AIApr 22, 2026·2 min 02Google releases offline dictation app for iOS powered by GemmaApr 22, 2026·2 min 03Former OpenAI Staff Quietly Launch Zero Shot, a Venture Fund Targeting $100MApr 22, 2026·2 min 04Startup Rocket Offers McKinsey-Style AI Consulting at a Fraction of the Usual CostApr 22, 2026·2 min 05Google updates Gemini with one-tap access to crisis mental health supportApr 22, 2026·2 min

Top themes

Startups·TechCrunch

Private capital is backing AI startups directly, bypassing venture funds

Robotics·The Verge

Intel to help build Terafab, an AI chip fab for SpaceX and Tesla

Creative·The Verge

Universal Music and Sony Refuse to Let Suno Distribute AI Tracks Outside Its App

Creative·TechCrunch

Google Maps can now write captions for your photos and videos with Gemini AI

Anthropic expands compute deal with Google and Broadcom as run-rate revenue hits $30 billion

Anthropic has expanded its compute agreement with Google and Broadcom as the company reports run-rate revenue of $30 billion a year.

Uber expands AWS contract, betting on Amazon chips over Google and Oracle

Uber is expanding its partnership with Amazon Web Services and moving key ride-hailing AI workloads to custom Trainium and Inferentia chips, choosing AWS over…

Hardware·The Verge

Anthropic launches Project Glasswing: new model finds vulnerabilities in every major OS and browser

Anthropic unveils Claude Mythos Preview, a closed cybersecurity model, in partnership with Nvidia, Google, AWS, Apple and Microsoft.

Ethics·TechCrunch

Anthropic releases Mythos preview, an AI model for defensive cybersecurity

Intel officially joins Elon Musk's Terafab chip project

Nvidia-backed AI data center builder Firmus hits $5.5 billion valuation

Apr 22, 2026·3 min

LLM·Wired

Anthropic Brings Apple, Google and 45 Other Organizations Together to Defend AI Against Cyberattacks

Creative·The Verge

Spotify adds AI podcast playlists built from a text prompt

LLM

Arcee, a 26-person startup, built a powerful open-source LLM and is winning users

LLM

Google quietly launches an AI dictation app for iOS that works without internet

Ethics

ProPublica Newsroom Strikes Over AI, Layoff Protections and Pay

Autonomous·TechCrunch

Atlassian adds visual AI and agents from Lovable, Replit and Gamma to Confluence

Confluence gains built-in tools for creating visual content and integration with AI agents from third-party services.

LLM·The Verge

Meta Superintelligence Labs Releases Its First AI Model, Muse Spark, for WhatsApp and Instagram

Meta Superintelligence Labs has launched Muse Spark, the first model in a new series. It already runs in Meta AI and is coming soon to WhatsApp, Instagram, Facebook…

Databricks co-founder Matei Zaharia wins ACM's top award, says AGI has already arrived

The creator of Apache Spark and Databricks co-founder has won the ACM's highest honor. He also says AGI is already here; people just misunderstand the term.

Research·TechCrunch

OpenAI publishes plan to protect children from AI-enabled sexual exploitation

OpenAI has released its Child Safety Blueprint, a document with four lines of effort against the use of generative AI to create child sexual exploitation material.