Sponsored by
Welcome Automaters, 👋
Congress really said, "Let's use AI to write actual federal laws," and honestly? We probably should've seen this mess coming from a mile away.
Here’s the absolute tea. According to a fresh report from Politico, an increasing number of members of Congress are out here letting tools like ChatGPT and Claude draft their official bills.
The main issue? The AI-generated bills are coming out so messy that the poor lawyers responsible for cleaning them up are practically drowning in slop.
The U.S. House Office of Legislative Counsel (a.k.a. the actual legal brain trust meant to turn wild ideas into legit legal language) is reportedly buried under a massive mountain of AI drafts.
We’re talking about drafts absolutely packed with:
Incorrect terminology that makes no legal sense
Mixed-up citations pointing to the wrong legal codes
Bizarre slip-ups that completely break the bill's logic
And here’s the absolute kicker: fixing these AI-generated nightmares is actually taking longer than if the lawyers had just written the bills themselves from scratch. Yikes.
Why should we care? Because federal law is definitely not the place for a casual "close enough" vibe.
Wade Ballou, who ran the Legislative Counsel office for nearly a decade until 2024, dropped some serious truth bombs on how AI completely misses the mark, and according to him:
AI consistently struggles to pick the right legal label for funding. It constantly mixes up tax credits, deductions, and grants. A small difference on screen, but a catastrophic difference for your wallet.
Even wilder? Certain AI models define "state" as strictly the 50 U.S. states. That tiny oversight means places like Washington, D.C., or tribal nations could accidentally be cut off from crucial federal programs overnight.
There’s also a much sneakier, far more alarming problem brewing behind closed doors.
Congressional staffers who rely on AI to draft these bills aren't spending time actually learning the nuances of the issues anymore. Basically, AI isn't just taking over the typing; it’s taking over the thinking. And that should terrify anyone who likes effective government.
The plot twist? The exact same office currently fighting off this AI chaos is now building its very own tool called the "Comparative Print Suite."
Instead of hallucinating wild legal theories, this internal tool is designed to:
Responsibly track how a new bill changes existing laws.
Flag human errors automatically instead of guessing random answers.
The Moral of the Story: AI can help document history, but maybe let's not give it the keys to the legal system unsupervised just yet.
One last thing before you move on: we just launched a new membership offering, and to celebrate it, we’re offering a 20% off discount for annual plans, but you have just 1 day to grab it. So go take a peek.
Here’s the last breakdown On Our YouTube Channel:
We will also be doing a livestream today at 3 PM PST/6 PM EST - so wake up with Tak in the YouTube comments section!
Here's what we have for you today
🤓 ChatGPT Teen Mode Rolls Out Globally: How OpenAI’s New Study Mode and Safety Limits Work
On August 18, OpenAI officially began rolling out ChatGPT for Teens worldwide. Think of it as a brand-new, heavily supervised version of the app built specifically for younger users.
And get this: you don't even have to sign up for it.
If the system suspects you are under 18; or if you've already admitted it; it automatically switches your profile over. It feels a little sneaky, but honestly? It’s for the best this time.
Wait, Why Now?
This major shift didn't just happen out of nowhere. OpenAI has been facing massive public pressure and legal heat following serious, tragic incidents involving teenagers using AI chatbots.
This rollout is essentially OpenAI putting its hands up and saying, "Okay, fine, we hear you! We are actually fixing this now."
So, What’s Actually Changing?
The upgrade breaks down into two huge pillars: Learning and Safety.
1. The Learning Side (a.k.a. Homework Hell for Teens)
Study Mode: Remember that feature from a few months ago, where ChatGPT won't just dump final answers on your screen, but instead asks guiding questions to force students to learn the material? Well, it's the feature that'll do the most for this new Teen Mode. In short, it's basically that one classmate who refuses to let you copy their homework.
Study Hours: Parents and teens can set specific time blocks where Study Mode stays turned on by default.
Quizzes & Visuals: The app now supports custom quizzes and interactive visual aids to help break down tricky topics, spanning above 300 different topics, with subjects like integrals, mitosis and moon phases covered.
Here is the funniest part! If the AI catches you trying to play the system to get quick answers, it will literally force you back into Study Mode. Absolutely iconic.
2. The Safety Side
No More AI "Besties": OpenAI is strictly tightening emotional boundaries. The AI is now forbidden from calling itself your friend or hinting that it has human feelings.
Quiet Hours & Break Reminders: Built-in cool-down periods so teens aren't doom-scrolling with their AI late at night.
Expanded Parent Alerts: OpenAI previously flagged accounts when a teen got kicked off for policy violations regarding violent threats. Now, linked parental accounts will receive alerts for discussions indicating serious self-harm concerns, including warning signs of eating disorders.
The Big Catch: Can Teens Hack It?
While OpenAI spokesperson Jonas admitted that “safety work is never finished” and that teen safety requires constant 24/7 updates, everyone is asking the exact same question: How long until teens completely outsmart this?
Let’s be real for a second. Teenagers are absolute geniuses at bypassing parental controls and digital locks. We have already watched platforms like Discord struggle to keep young users from dodging age restrictions.
Until this system goes through real-world stress testing, nobody actually knows how easy it will be for a clever teen to escape these new rules. And with all the current regulatory eyes watching, OpenAI simply cannot afford a massive security leak here.
The Rollout Plan: The global update started today and is expected to reach every user within two weeks.
The Bottom Line: ChatGPT just grew up a little bit, all so it can help teens grow up safely too.
P.S. We're dissecting all of this over on our YouTube channel! Head over to The Automated, hit that subscribe button, and show up in the comment section with Tak! You can drop a comment letting us know which section you liked best.
Oh, and mark your calendars: new drops land 6 AM Shenzhen / 2 PM EST or 5 PM EST, Tuesday through Saturday China time (Monday to Friday U.S.A time). Come catch us live!
The Voice Data Behind the World's Best AI
Training data quality determines model quality. Voices delivers custom or pre-built voice datasets—directed by real professional talent, not scraped audio—across any language, accent, or emotional range.
Choose from 800+ hours of character voice performances, 1,723 hours of QA'd conversational data across 56 emotional states, or 1,000 hours of expressive, multi-language data with full emotion tagging. Every dataset ships with clear licensing, signed consent, and usage rights baked in, so your legal team sleeps easy and your model trains on data it can actually use.
7 of the world's top 10 AI labs already use Voices. Request free samples and hear the difference professionally directed voice data makes.
🧱 Around The AI Block
🙅♂️ OpenAI denies disbanding its preparedness team.
👩💻 Cursor capitalizes on GitHub frustration, and launches rival hosting platform.
🛡️ OpenAI institutes new safeguards after Hugging Face breach.
🦾 Google rolls out Gemini AI assistant in Chrome for all U.S. Android users.
🤖 OpenAI announces slowing pace of development after hack by rogue agent.
✍️ Pennsylvania governor signs order imposing new rules to set up AI data centers in state.
📦 Nvidia H200 chips reach China in small shipments.
🦿 Warp’s new system is an out-of-the-box software factory for AI development.
🛠️ Trending Tools
For the AI Engineers & System Reliability Teams: Diagrid ensures high reliability for autonomous agents and workflows by offering state preservation, automatic failure recovery, and step-by-step action tracking. It features cryptographic audit logs to trace every interaction, making complex multi-step systems easy to inspect and debug. Compatible with popular frameworks like LangGraph, CrewAI, and Google ADK, it gives developers a dependable production backbone for mission-critical tasks.
For the Sales Leaders & B2B Revenue Teams: Trayo AI monitors target accounts in real time to automatically catch high-intent buying signals the moment they occur. It filters out irrelevant market noise, enriches actionable leads with context, and highlights the precise moment to reach out to potential buyers. It gives sales teams a distinct timing advantage by turning quiet prospect activity into warm pipeline opportunities.
For the Style Consultants & Personal Care Creators: KnowMyFace analyzes facial photos to map precise structural geometry, prominent features, apparent age, and visible skin metrics. It converts these visual measurements into actionable style insights, including personalized makeup techniques and haircut recommendations tailored to individual face shapes. It offers an easy way for personal care brands or individuals to discover flattering beauty choices guided by computer vision.
🤖 AI Workout Of The Day: Best Ways to Use Google NotebookLM
Alright, quick PSA: If Google’s NotebookLM is still sitting untouched in your bookmarks… we need to talk.
Because this thing? Quietly goated.
NotebookLM isn’t just another “ask me anything” AI. It’s more like a brainy co-pilot that only cares about your stuff. You drop in your docs, links, videos, audio, whatever, and it thinks inside that universe. Like seriously, no random hallucinations. Just grounded, cite-everything energy, powered by Gemini.
Why it slaps:
Auto summaries that don’t miss: Upload sources and boom, clean summaries, key themes, inline citations. Research without the headache.
Source-locked answers: Every response is pulled directly from your materials, with receipts. If it says it, it can point to where it came from.
AI podcast mode (yes, really): It spins your content into a downloadable audio convo between two AI hosts. You can even tell them what angle to take. I mean genius right?
Getting started (takes like 2 minutes):
Sign in
Create a notebook per project
Upload PDFs, Docs, Slides, URLs, YouTube transcripts, audio files, or straight-up pasted text
Pro moves:
Ask tight, specific questions
Use the "save to note" button to preserve great answers.
End sessions by asking for a convo summary
Experiment with formats (timelines, pros/cons, comparisons)
Feed it good sources—cuz garbage in, garbage out
Use the Notebook Guide for instant templates and starter questions
Big picture? NotebookLM doesn’t try to replace your thinking, it sharpens it. If you work with research, content, or complex ideas and you’re not using this yet… respectfully, what are we doing?
Go try it.
PS: We've got the tutorials you need to kill it in your next project:
💡 Prompts To Try For Connecting Concepts:
You are a learning strategist. Help me understand how concepts connect across the sources I provided:
Core concept: [MAIN TOPIC]
Related project: [WHAT YOU'RE BUILDING]
Using my uploaded sources, create:
1. Mind map of related concepts and their relationships
2. Examples of how this concept applies in different contexts
3. Prerequisites I should understand first
4. Common misconceptions to avoid
5. Real-world examples that demonstrate the conceptIs this your AI Workout of the Week (WoW)? Cast your vote!
That's all we've got for you today.
Did you like today's content? We'd love to hear from you! Please share your thoughts on our content below👇
What'd you think of today's email?
Your feedback means a lot to us and helps improve the quality of our newsletter.
