In partnership with
Welcome back, AI enthusiast.
Elon Musk took the stand in his lawsuit against OpenAI and things turned messy fast.
His own tweets were used to contradict him under oath. Can you trust a man who argues with himself online?
Today in the “Incredible AI” world:
Musk’s Own Posts Betrayed Him in Court
IBM's New Speech Models Are Impressive
Running AI Evals Is Draining Budgets Now
Simplify Employee Time Tracking with AI
5 Super-Useful AI Tools You Should Try
Top AI-Generated Image of the Day
Read Time: 4 minutes
- LATEST DEVELOPMENTS -
Musk's Own Tweets Put Him on Trial:
Elon Musk walked into a California courtroom to accuse Sam Altman of stealing a charity. He walked out having contradicted his own tweets under oath. It was not a great day for the world's richest man.
Things to Know:
The Core Accusation: Musk says Altman tricked him into funding a nonprofit, then quietly turned it into a money machine. He called the whole thing "looting the nonprofit" in dramatic fashion.
The Tweet That Bit Back: After claiming Tesla isn't chasing AGI in court, lawyers pulled up his own post saying Tesla "will be one of the companies to make AGI." Awkward silence followed.
The Money Problem: Musk pledged $1 billion to OpenAI but only wired $38 million. His defense? His reputation made up the difference. The courtroom did not burst into applause.
Microsoft as the Villain: Musk says Microsoft's $10 billion investment was the moment OpenAI stopped being a charity and started being a business, calling it a clear "bait and switch."
Here's the thing about suing someone while your own tweets are used as evidence against you: it does not go smoothly. Musk returns today for more cross-examination, and more X posts are probably waiting in the queue. Grab some popcorn.
IBM Pushes Granite Speech Models Forward:
IBM just released two new open-source speech recognition models under the Granite 4.1 family. Both are free to use under the Apache 2.0 license and already sitting on Hugging Face. Small size, serious performance.
Things to Know:
Two Models, Two Missions: The standard Granite Speech 4.1 2B handles transcription and translation across six languages. The NAR variant skips translation but delivers much faster results for teams that need speed over features.
The Language Lineup: Both models cover English, French, German, Spanish, and Portuguese. Only the standard model adds Japanese and speech translation, so pick your version based on what your project needs.
The Speed Stat That Stands Out: The NAR model processes audio roughly 1,820 times faster than real time on a single H100 GPU. A one-hour recording done in under two seconds is the kind of number that makes engineers smile.
IBM quietly slipped in a bonus model called Granite Speech 4.1 2B-Plus. It adds speaker identification and word-level timestamps, which is useful when you need to know who said what and exactly when.
Your next great hire lives in Slack.
Viktor is an AI coworker that connects to your tools and ships real work. Ask Viktor to pull a report, build a client dashboard, or source 200 leads matching your ICP. Most teams hand over half their ops within a week.
AI Benchmarks Testing are Getting too Costly:
Testing whether an AI model actually works used to be the cheap part of the job. That is no longer true. One benchmark run can now cost more than training the model you are testing.
Things to Know:
The Numbers Are Wild: The Holistic Agent Leaderboard spent $40,000 to run just 9 models across 9 benchmarks. A single GAIA benchmark run on a frontier model can hit $2,829 before any caching tricks.
Agents Make Everything Worse: Static benchmarks can be compressed 100 to 200 times without losing ranking accuracy. Agent benchmarks? You get maybe 2 to 3.5 times compression at best. The math just does not work the same way.
Reliability Costs Even More: Running a benchmark once tells you almost nothing solid. Adding 8 repeated runs for statistical reliability multiplies the $40,000 HAL sweep to roughly $320,000. Most teams cannot afford that.
Who Gets Locked Out: When frontier evaluations cost tens of thousands per run, only the biggest labs can afford them. Academic groups and regulators get priced out, and the labs end up grading their own homework.
The fix is not purely technical. It is about sharing eval data so every team stops re-running identical experiments from scratch. One reuse of a $9,500 benchmark saves more than most compression techniques ever could.
- TRY THIS -
How to Track Employee Hours Automatically:
If you manage a team of hourly workers, you know how much time disappears into chasing timesheets, fixing errors, and manually calculating overtime before every payroll run.
Buddy Punch is a time tracking and scheduling tool built for small businesses with hourly employees. It replaces paper timesheets and manual processes with an automated system that feeds directly into payroll.
How it works:
Add your employees: Sign up at buddypunch.com and enter your employee list individually or in bulk. They don't need company email addresses to be added as users.
Let employees clock in and out: Employees punch in via browser, mobile app, or a kiosk using a PIN, QR code, or facial recognition. GPS tracking and geofencing ensure they're actually on-site when they do.
Timesheets fill out automatically: Hours, breaks, and overtime are calculated without anyone doing the math manually. You can require manager approval before timesheets are finalized.
Run payroll directly or export: Connect Buddy Punch to QuickBooks, Gusto, or ADP, or use the built-in payroll feature to process everything in one place.
If payroll currently eats up half your Friday, Buddy Punch is worth a look.
The best prompt engineers aren't typing. They're talking.
Power users figured this out early: speaking a prompt gives you 10x more context in half the time. You include the edge cases, the examples, the tone you want — because talking is fast enough that you don't skip them.
Wispr Flow captures everything you say and turns it into clean, structured text for any AI tool. Speak messy. Get polished input. Paste into ChatGPT, Claude, Cursor, or wherever you work.
89% of messages sent with zero edits. 4x faster than typing. Works system-wide on Mac, Windows, and iPhone.
- BEST AI TOOLS -
Fliki: Convert text into videos with realistic AI voices and dynamic visuals.
Consensus: Search research papers and extract scientific answers using AI analysis.
Nureply: Automate cold email campaigns with AI-personalized outreach.
Upscayl: Enhance and upscale images using open-source AI models.
Recast AI: Transform articles into conversational AI voice summaries.
- DAILY POLL AND RESULT -
Today’s Poll:
Q) Is AI-generated humor actually funny?
Vote and find out the result tomorrow.
Yesterday’s Result:
Q) Should content creators disclose AI voiceovers?
A) Yes, audience deserves clarity - 66% 👑
B) No, output is what matters - 44%
The Best New App for Reading Newsletters:
Reading newsletters in the inbox is frustrating - it is noisy and easy to lose control of subscriptions.
That’s where Meco comes in as a savior.
Meco is a distraction-free app for reading newsletters outside the inbox. The app is packed with features designed to supercharge your learning from your favorite writers.
With Meco, you can enjoy your newsletters in an app built for reading, while giving your inbox space to breathe.
Add your newsletters in seconds and liberate your inbox today!
- AI IMAGE OF THE DAY -
Mustard Gas Chaos Engulfs Battlefield Soldiers:
Prompt:
"A gritty WW1 battlefield scene where exhausted soldiers frantically adjust their gas masks in the trenches. Surrounding them, giant Heinz mustard bottles have shattered on the ground, releasing thick yellow smoke. More bottles arc through the hazy sky mid-fall, as pale light filters through the toxic fog casting long shadows across the mud."Try this prompt in any decent AI image generator and let me know your result.
- DO YOURSELF A FAVOUR -
If this email landed in your Promotions or Spam tab, move it to your Primary inbox. Takes 2 seconds, and you will never miss an edition again.
I put this together every day so you do not have to spend your morning hunting down AI news yourself. Would be a shame if it ended up buried under discount codes and bank alerts.
