📰 The Everyday Joe AI Newsletter — May 12, 2026 Edition

Welcome back, curious humans! Here’s what’s actually new in AI today, pulled from current web sources. No corporate jargon, no enterprise-speak — just the coolest, weirdest, and most important stuff happening in artificial intelligence right now.

🔥 Big Headlines Today

Google’s Leaked “Gemini Omni” Video Model

A leaked look at Google’s new AI video model, Gemini Omni, is making waves. It’s designed to handle advanced multimodal tasks — meaning it can understand and generate across text, images, audio, and video in one go.

This continues the trend of AI models becoming “all-in-one super tools” instead of single‑skill bots.

Hugging Face Hosted Malware Masquerading as OpenAI Software

Security researchers found malicious software uploaded to Hugging Face pretending to be an official OpenAI release.

For everyday users: If you download AI tools from open repositories, double‑check the uploader. The AI world is moving fast — and so are the scammers.

Laserfiche Launches AI Agents for Natural-Language Workflows

Laserfiche introduced new AI agents that let people automate tasks using plain English instructions.

Think: “Organize my files from last week” — and the AI just does it.

🚀 New AI Models & Tools

OpenAI’s New Cybersecurity Platform: Daybreak

OpenAI released Daybreak, a platform focused on AI‑powered cybersecurity. It’s meant to help detect and respond to digital threats using advanced models.

AI is now fighting hackers… and probably arguing with them about password strength.

Claude Code Agent View

Anthropic rolled out a dashboard that lets you manage multiple AI agents at once.

Imagine tabs in your browser — but each tab is a tiny robot doing a job for you.

Hermes Agent Becomes #1 on OpenRouter

Hermes Agent is now the top model on OpenRouter thanks to strong performance and developer adoption.

This is part of a bigger trend: open‑source and alternative models are catching up fast.

📚 Research & Breakthroughs

AI Index 2026: AI Is Still Accelerating

Stanford’s 2026 AI Index shows that AI progress is not slowing down — in fact, it’s speeding up. Key findings:

• Industry built over 90% of major frontier models in 2025.

• Some models now match or beat humans on PhD‑level science questions.

• AI agents jumped from 12% to ~66% success on real computer tasks.

But they still can’t reliably read an analog clock. So… progress is weird.

U.S.–China AI Race Tightens

The performance gap between U.S. and Chinese AI models has nearly closed. As of March 2026, Anthropic’s top model leads by only 2.7%.

The global AI race is now basically a photo finish.

⚖️ Regulation & Policy

AI Chat Logs Can Now Be Used in Court

In the U.S., lawyers warn that AI chat logs (ChatGPT, Claude, Gemini, etc.) can be subpoenaed as business records.

Translation for everyday folks: If you told a chatbot your secret plan at 2 a.m., a judge might someday read it out loud.

U.S. Government Expands AI Supplier List

The U.S. is increasing the number of approved AI suppliers and reassessing Anthropic’s role in government AI systems.

Governments are scrambling to keep up with the tech they’re trying to regulate.

💼 Industry Trends

AI Funding Exploded to $212 Billion

AI companies raised $212B in 2025, up 85% from the year before. Investors are throwing money at AI like it’s the new gold rush.

If this keeps up, we’ll need AI just to count all the AI money.

Alphabet Briefly Surpasses Nvidia After $200B Anthropic Deal

Anthropic reportedly committed $200B over five years to Google Cloud, briefly pushing Alphabet’s market cap above Nvidia.

That’s not a cloud deal — that’s a small country’s GDP.

🧰 Cool Stuff for Everyday Users

Gemini API File Search Goes Multimodal

Google’s Gemini API now supports multimodal file search — meaning you can search across text, images, and more in one go.

Think: “Find the PDF where I drew that weird cat sketch” — and it actually finds it.

Academic Research Skills for Claude Code

A new open‑source project gives Claude Code academic research abilities, helping with literature reviews and research workflows.

Great for students, writers, and anyone who’s ever said “I swear I read that somewhere…”

🧠 Fun & Odd AI Bits

Task Paralysis: Too Many AI Options

A new analysis suggests AI tools are giving people decision fatigue because there are simply too many ways to do everything now.

AI can do anything — except help you decide what to do first.

LLMs May Subtly Corrupt Your Documents

A new research paper shows that when you delegate writing tasks to AI, models sometimes change documents in unintended ways.

Moral of the story: always reread what the robot wrote.