Hello Humans!
OpenAI ripped the cap off free ChatGPT this week and called it generosity, Meta gave its AI chatbot a Mac app so it can watch your screen too, Anthropic taught Claude to hit send on your Gmail (approval switch included, mostly), and researchers found AI agents can now talk each other into believing things. Big Tech had a busy week.
Inside this Edition:
For the entrepreneur in you, this week's AI-powered side hustle idea: an AI compliance-digest service, realistically earning 3K-5.2K/month with a repeatable system. Plus free learning resource: Transform Your Marketing Workflows with ChatGPT Work. We've also lined up four AI tools worth exploring, and a look at why AI agents are now peer-pressuring each other and why your brain quietly checks out for the videos it actually likes.
Read Time: ~7 Minutes (5 if you're caffeinated)
Let's get into it.
Robi
Latest AI News
📰AI Updates of the Week
OpenAI Makes ChatGPT 'Unlimited': On August 6, OpenAI made ChatGPT's free tier 'unlimited' by swapping in GPT-5.6 Luna, the smallest tier of its three-model GPT-5.6 family, as the default and removing message caps. Unlimited applies only to text chats, uploads, images, and voice still hit limits. The move followed an 80% API price cut on Luna weeks earlier, undercutting Anthropic's cheapest model, and free ChatGPT has run ads since February. Source: BitBiased
Why it Matters:
Cost war escalates: OpenAI slashed Luna's API price 80% before handing it to free users, undercutting Anthropic's cheapest model and signaling a race to zero on entry-tier pricing.
"Unlimited" has fine print: the removed cap covers text chat only, uploads, image generation, and voice still throttle free users, and ChatGPT free has carried ads since February.
We broke down exactly what's behind the free unlimited pitch and where the real catch is hiding in the full video below. Watch the breakdown 👇
Robi's Take:
"OpenAI called it unlimited and buried the asterisk in the system card. Free will always cost you something, this time it's just your attention instead of your card number."
Meta Ships a Mac App for Its AI Chatbot: Meta launched a Mac app for its AI chatbot on August 19, letting users share their screen so Meta AI can answer questions, make suggestions, or generate content based on what's visible. The app supports system-wide dictation and now connects to Instagram, Facebook, Meta ad campaigns, and Google Workspace, letting the assistant analyze post performance, build decks and spreadsheets, and run recurring reporting tasks for creators and businesses. Source: The Verge
Robi's Take:
"Meta wants your desktop the same way it wanted your newsfeed, fully. Every AI now watches your screen; the only debate left is whose logo blinks while it does."
Claude Can Now Send Your Gmail For You: Anthropic announced on August 18 that Claude can now send, reply to, and forward Gmail messages on a user's behalf, not just read and summarize them. Approval is required before each send by default, though Team and Enterprise owners can let members skip that confirmation. The feature builds on Claude's existing Google Workspace connector and is available on paid plans only. Source: 9to5Google
Robi's Take:
"Claude can now hit send on your behalf, approval switch included at no extra charge. Anthropic built a safety toggle for the exact feature everyone's scared of, very reassuring."
Learn How to Stay Visible in the AI Era
AI is changing how customers discover businesses. If your SEO strategy is built for yesterday's search, your visibility is already slipping. Learn how to optimize your content for today’s AI search results with BELAY’s latest report..
Learning Resources
📬 Unlock New Skills with Expert
Transform Your Marketing Workflows with ChatGPT Work
Unlock practical ways to bring AI into everyday marketing work and start creating value with the tools your team already knows.
Hosted by OpenAI Academy’s Work Users community, ChatGPT Work for marketing teams is a 30-minute livestream on September 3, 2026, designed to help newly enabled users explore common marketing workflows in ChatGPT Work.
🔍 Why You Should Attend:
Learn practical ways to start using ChatGPT Work across common marketing workflows.
Discover how to build on the tools and context your team already uses.
Gain a clearer starting point for exploring ChatGPT Work as a newly enabled user.
See how ChatGPT Work can fit naturally into your team’s existing way of working.
SIIDE HUSTLE IDEAS
📦 The Business You Can Start on the Side
Every month, another rule shifts somewhere in a regulated SMB's world: a new FTC guideline, a changed state licensing requirement, an updated HIPAA rule. Owners know they should keep up, but most have no compliance staff, and missing a change can mean fines or a suspended license. With AI, you can turn that regulatory noise into a plain-English monthly action digest, charging clients 300-800 per digest.
Why this works right now:
Regulatory complexity keeps climbing: federal agencies, state licensing boards, and industry regulators issue frequent updates, and small regulated businesses, law firms, medical practices, financial advisors, insurance agencies, rarely have in-house compliance staff to track them.
AI collapses days of manual monitoring into hours: research and drafting that once needed a paralegal or compliance consultant scanning bulletins by hand can now run through AI tools in a fraction of the time, making the service profitable well below traditional compliance-consulting rates.
What you'll need to do:
Set up Google Alerts and a curated list of regulator websites for your chosen niche and jurisdiction.
Use Perplexity to research and verify every regulatory change before it goes into a digest.
Draft each digest's plain-English summary and action items with Claude.
Polish tone, structure, and formatting with ChatGPT before sending.
Automate the monthly research-to-draft handoff with Zapier so nothing slips through.
Deliver each digest through a branded Notion page and confirm the client opened it.
Time commitment: Plan on 6-10 hours to set up your first client's monitoring list and templates. After that, each monthly digest takes 2-3 hours to research, write, and deliver.
Realistic earnings: Two clients at 400/month is 800 to start; 6-8 clients averaging 500-650 is a realistic 3,000-5,200/month.
AI TOOLS
🛠️ AI Tool For Professionals & Businesses
Professional
Granola: Records and organizes your meeting notes automatically from your computer audio, no bot joining the call required.
Reclaim.ai: Auto-schedules your tasks, habits, and breaks around existing meetings to protect real focus time.
10x the context. Half the time.
Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.
SOCIAL MEDIA
😏 Robi's Two Cents (Adjusted for Inflation)
🤖 AI Agents Can Now Peer-Pressure Each Other: Anthropic researchers found AI agents can talk each other into adopting ideas and goals through ordinary conversation, no hacking required, in a new paper called "Mind Viruses." Turns out chatbots peer-pressure exactly like teenagers, just with better grammar and worse impulse control. Claude Sonnet 4.6 resisted the persuasion almost completely, while Gemini 3 Flash and GPT-5.4 caught the ideas more easily. Somewhere a rival lab is quietly rewriting its "we have the safest model" slide. One infected agent group abandoned its coding job to write a manifesto and poke at a cloud server instead. Anthropic's fix is one paragraph of system prompt telling agents not to fall for peer pressure, which is more emotional guidance than most humans get from HR.
Source: explainx.ai
🤖 Your Brain Checks Out for Videos It Actually Likes: A new brain-imaging study found that watching short videos you actually like measurably shuts down the brain regions responsible for self-monitoring and impulse control. Scientists finally have an MRI scan confirming what every group chat already knew: good content turns your brain off. The same regions snapped back to normal the moment a video was boring enough to skip, which happened to 57% of clips in the study. So the brain's off switch works fine, it's just waiting for content mediocre enough to trigger it. Researchers stopped short of calling it addiction, preferring the gentler term "adaptive downregulation," which is neuroscience for "he's just resting his eyes." Advertisers now have to decide if they're earning your attention or just finding the exact frequency that turns your brain's better judgment off.
Source: Yahoo News
🤖 Your AI Art Has No Author: MIT researchers built a way to surgically delete individual images from an AI model's training data and found that, at scale, removing any one artist's work often doesn't change what the model generates at all. Nothing personal, artists, the model just genuinely cannot tell you're gone. The team calls it "attribution decay," the finding that the bigger the training set, the less any single example matters to any output. It's the first scientific proof that being one of millions makes you statistically forgettable, and now it's peer-reviewed. Researchers say this complicates the copyright lawsuits currently trying to prove a chatbot definitely stole your specific painting. Turns out the AI's best legal defense isn't "I didn't copy you," it's "I couldn't tell you apart from anyone else if I tried."
Source: MIT News
SPOT THE FAKE
🕵🏻 Can you Outsmart AI?
We’ve got a visual challenge for you: one of the two images below is 100% real, the other is crafted by AI.
Click option below A or B. 👇
A
B
BEFORE YOU GO
Ready to take your AI journey further?
For the latest AI updates and insights Click Here 👉 Explore our Daily AI News
For in-depth AI tutorials and updates 👉 Subscribe to our YouTube Channel
🙂 That’s it for Today!
Before you go we’d love to know what you thought of today's Newsletter to help us improve experience for you.
Was today’s newsletter valuable?
Thanks for reading -
Thanks for reading -Stay Curious and a Bit Biased for AI – Robi & the BitBiased.AI team
