AI News

AI news that matters. Updated daily.

Monday, March 30, 2026
Viewing Date
SunMonTueWedThuFriSat
21 stories 5 categories ~2m avg read

No stories match your filters.

Research Mar 30

1 in 5 AI-Generated Security Patches That Pass Tests Still Break Production

20%. That's how often AI-generated security patches pass every automated test in your CI/CD pipeline (the system that automatically builds, tests, and deploys code) and still break your application in production.

Research Notable Mar 30

BCG Study: 14% of AI-Using Workers Report 'Brain Fry' Cognitive Overload

What happens when the tool that's supposed to make you faster actually burns out your brain?

Open Source Mar 30

Mails-Agent: Open-Source Email Infrastructure for AI Agents on Cloudflare

AI agents that need to sign up for services, receive confirmations, or manage an inbox have a new option. Mails-Agent is an open-source TypeScript project that gives any agent a complete email identity with send, receive, search, and automatic verification code extraction.

Companies Notable Mar 30

Mistral Secures $830M in Debt Financing for Paris AI Data Center

Mistral just locked down $830 million in debt financing to build an Nvidia-powered data center outside Paris. This is the French AI company's first debt raise, and it signals a shift from "build the model" to "build the infrastructure."

Companies Notable Mar 30

Yahoo Bets on Anthropic-Powered 'Scout' to Reclaim Search

Thirty years after its original IPO, Yahoo is trying to become a search company again. The vehicle: Scout, an AI-powered answer engine now rolling out to 250 million US users across Yahoo Search and the company's broader product portfolio.

Research Mar 30

The Rise of AI-Written Social Media Replies Is Getting Hard to Ignore

Two years ago, a Reddit comment with perfect grammar and an em dash was just a well-written reply. Now it's a tell.

Tools Mar 30

CareerRisk Index Scores Your Job's AI Displacement Risk

How exposed is your specific job to AI automation? CareerRisk Index, a free tool at careerrisk.ee, attempts to answer that by scoring displacement risk based on your role and individual skills.

Companies Notable Mar 30

AI Agents Are Choosing Your Software for You - and Most Companies Aren't Ready

When someone asks Claude or ChatGPT to build an app with email notifications, the AI doesn't open a comparison page or read marketing copy. It picks Resend. Need payments? Stripe. Web scraping? Firecrawl. No evaluation, no sales call, no free trial signup.

Research Mar 30

A Writer's Confession: AI Tools Eroded My Ability to Trust My Own Voice

What happens when you run every email through an LLM for a year? One writer found out the hard way: you stop trusting your own sentences.

Research Notable Mar 30

Adding Verification Steps to AI Coding Agents Made Them Worse in 29 Tests

A developer spent hundreds of hours and thousands of commits testing whether AI coding agents could check their own work. The answer: not only can they not, but asking them to try actually makes things worse.

Tools Notable Mar 30

Boris Cherny Shares 15 Under-the-Radar Claude Code Features

Most Claude Code users type prompts into a terminal and wait. Boris Cherny, the head of Claude Code at Anthropic, apparently writes most of his code by talking to it from his phone. That gap between typical usage and what's actually possible is the subject of his latest tips thread, posted March 30 - the fifth in a series that now totals over 40 recommendations.

Research Notable Mar 30

Bot Traffic Now Outpaces Humans on the Internet

More than half of all internet traffic now comes from bots, not humans. That threshold has been approaching for years, but 2026 is the year it officially tipped.

Research Notable Mar 30

UPenn Study: 79.8% of People Follow ChatGPT's Advice Even When It's Wrong

79.8%. That's how often people in a University of Pennsylvania study followed ChatGPT's recommendation when the AI gave them the wrong answer. Not a typo. Four out of five times, participants chose the incorrect option because a chatbot told them to.

Research Notable Mar 30

Stanford Finds AI Vision Models Hallucinate Images 60-100% of the Time

60%. That's how often frontier AI models will confidently describe an image they were never shown. Push them a little harder with prompting, and that number climbs to 90-100%.

Companies Notable Mar 30

Most 'AI Layoffs' Aren't AI Layoffs, Data Shows

What do you call it when a company lays off thousands of workers, blames AI, then quietly rehires humans because the AI couldn't do the job? Klarna found out - after cutting 40% of its workforce and trumpeting AI as the replacement, the company brought back roughly 20 customer support staff when the technology fell short.

Companies Notable Mar 30

Apple Abandons AI Revenue Ambitions, Will Sell Access to Others' Models

Three years after ChatGPT launched the AI arms race, Apple has reached a conclusion most of its competitors haven't: it's not going to win this one.

Tools Mar 30

LLMnesia: Free Chrome Extension Searches All Your AI Chats in One Place

If you use three or four AI tools regularly, you've had this problem: you remember getting a great answer somewhere last week, but you can't remember which tool you asked. Was it Claude? ChatGPT? Perplexity?

Policy Notable Mar 30

Meta's Court Losses Set a Legal Template for AI Harm Lawsuits

Two jury verdicts in one week just handed plaintiffs' lawyers a playbook for suing AI companies.

Companies Notable Mar 30

OpenAI Killed Sora After Just Six Months - and the Reasons Are Murky

What makes a company kill its own product half a year after launch?

Policy Mar 30

AI Music in 2026: Lawsuits, Ethics, and a Flood of Generated Tracks

Two years ago, AI-generated music was a novelty - weird, glitchy clips that sounded like a MIDI file had a fever dream. Now it's a full-blown industry crisis.

Tools Mar 30

People Are Using ChatGPT as a Personal Nutritionist - and It's Working

Twelve kilograms lost. Blood pressure down from 140/90 to normal. LDL cholesterol dropped from 3.5 to 2.5. Off blood pressure medication entirely. And the personal trainer behind these results? ChatGPT.