In partnership with
Welcome Automaters, 👋
So guess what, that "harmless" holiday photo you just posted to your story? Yeah, it might be working overtime against your bank account.
Picture this: You are living your absolute best life in Spain. You snap a cute family picture with the river flowing gracefully in the background, post it, and call it a day. You didn't drop a location tag, so obviously nobody knows where you are... right?
Wrong. Apparently, AI knows exactly where you’re hanging out.
Here’s how the messy drama unfolds. A couple of days after your dream trip, your phone buzzes with a text that says: "We detected unusual activity on your card while you were traveling in Spain. Please verify immediately."
It feels completely legitimate, you panic, click the link, and end up handing over your bank details. Except you just got played!
And here’s the absolute scariest part: Criminals never needed your geotags or metadata. They simply used AI to analyze the actual photo, zooming in on background clues like architectural styles, signage, and even the subtle way light hits a building.
How Do the AI Detectives Spot You?
According to fresh research from cybersecurity team McAfee, (as reported by The Guardians) AI models were fed over 21,000 travel photos, and the accuracy was straight-up unsettling:
Model #1 correctly guessed exact locations 91% of the time.
Model #2 nailed it 87% of the time.
When staff members fed ChatGPT a random picture of a river with a few trees, the AI instantly pinpointed it as Hastings-on-Hudson in New York.
Another simple photo of tulips was immediately flagged as the famous Keukenhof Gardens in the Netherlands, purely because the AI recognized the specific way the blue flowers were planted between the tulips!
The McAfee team even ran tests on their own personal vacation snapshots and got completely spooked by how effortlessly their private travels were exposed.
And get this: When criminals know your location history, their fake outreach becomes terrifyingly believable. They might text you about suspicious charges in a specific city, claim to be calling about your recent hotel stay, or ask you to confirm a login attempt from the exact country you just visited.
As McAfee's EMEA lead, Vonny Gamot, put it: AI provides the missing context that makes these dangerous scams feel 100% credible.
So, how do we keep living our best lives online without handing scammers the keys to our financial kingdom? Here’s your game plan:
Delay the post. Wait until you are safely back home before sharing your vacation dump online.
Lock down your profile. Keep your social media accounts private so only trusted friends can view your moments.
Never panic-click. Scammers thrive on urgency. If a text demands you "act now," force yourself to slow down.
Verify on your own terms. Always skip the provided links and never click on those. If your bank supposedly texts you, call them back directly using the details on their website or on the back of your debit card.
Bottom line: Your photos are spilling all your secret tea, even when you aren't saying a word. It’s officially time to make them a little less chatty!
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, but you have just 3 days to grab it. So go take a peek.
Here’s the last breakdown:
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
😳Anthropic Explains How Claude's AI Watermarks Actually Work, While Google Lets You Turn Yours Off
Last week was basically a full-blown watermark soap opera. And this week, two of the biggest powerhouses in AI just dropped massive news about the secret (and not-so-secret) stamps hidden inside your content, and the vibes could not possibly be more different.
First up on the stage: Anthropic. After dropping the bombshell that they’re officially watermarking text produced by Claude to comply with the EU AI Act’s Transparency Code, the internet completely spiraled. People had questions, so Anthropic published a breakdown to clear the air.
Here’s the juicy part made super simple:
How it works: When Claude decides between two words that mean the exact same thing (think "grey" vs "overcast"), it subtly leans toward specific words in a predictable pattern.
The hidden trail: You cannot see the pattern with the naked eye, but anyone holding a special key can spot it instantly. Anthropic is basically using the SynthID-Text framework developed by Google DeepMind to leave digital breadcrumbs across your writing.
Detection vs Watermarking: Anthropic made sure to clarify that this is not like third-party AI detectors that just guess based on repetitive writing tropes. This is a deliberate, mathematically baked-in stamp!
Naturally, the internet reacted with zero chill! Over on Reddit, users were divided into two warring camps: one group called it a massive conspiracy against innocent creators, while someone else snapped back, "The only reason you wouldn’t want this is to lie to people."
Meanwhile, Business Insider reported that dozens of frustrated users on X claimed they canceled their Claude subscriptions over the change. Yikes! Anthropic tried to calm everyone down by stressing that responses will look completely identical to readers, but the panic was already in full swing.
So Can Claude’s Watermark be removed?
Light Edits / Proofreading won't erase it. If Claude heavily edits or rewrites your work, the watermark sticks like glue.
Rewriting the entire text yourself strips it away.
Code is barely impacted because programs need precise syntax to run, though comments might carry tiny watermark traces.
Anthropic also subtly hinted that Claude will not be the only snitch on the block soon, noting that every major model developer signed the exact same Code of Practice!
Meanwhile Google Steps In With The Ultimate Shaded Move
While Anthropic was busy fending off angry subscribers, Google casually entered the chat with the opposite energy.
They announced a brand-new setting that allows users to turn off visible watermarks on AI-generated photos, videos, and music, in countries that allow that. So EU citizens might still have to deal with the watermarks! However, the timing feels like a total power move, basically signaling to creators, "Hey, our platform actually gives you control."
Where to find it: Rolling out over the coming days, you can toggle this off under Settings > Media Watermark.
Supported tech: Josh Woodward, Google’s VP for Gemini, confirmed this covers Nano Banana, Omni, and Lyria models across Gemini and Flow, with Search coming next.
The fine print: This only removes the visible watermark overlay. The hidden, invisible SynthID tag and official C2PA metadata will stay permanently intact so people can still check for authenticity using Gemini or Search.
Developer bonus: Google is also open-sourcing a new library named Credentio to help devs build local validation checks directly into their own applications.
According to Google, visible logos can completely ruin a project when you’re trying to create polished, professional work, so Google is giving creators their aesthetic freedom back while keeping the safety checks hidden safely under the hood.
The Bottom Line:
Anthropic is locking things down on one side of the world to satisfy regulators, even if it pisses off loyal fans, while Google is handing creators the eraser to win over hearts and minds. In short, the industry is still wildly confused about where to draw the line between transparency, safety and usability, but one thing is clear: the AI watermark wars are just getting started!
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!
Elon's Launching His Own "iPhone." 3 Stocks to Buy to Prepare
If you'd bought Apple before the first iPhone, you'd have seen peak gains of 7,537%. One product turned ordinary investors into millionaires.
Now Elon Musk is launching his own version — a device he claims will be "10x bigger than the largest product in history."
Target launch: July 22. Just days away.
The people who win on launches like this don't buy after the headlines. They position before. Our analyst pinpointed 3 stocks set to ride the launch — with entry guidance and price targets, plus a bonus 4th pick (the most undervalued name in the supply chain) and a 3-phase playbook showing exactly when to buy, when to add, and when to take profits.
100% free. No credit card. Over 2,500 investors have already read it.
🧱 Around The AI Block
🙅♂️ Why people aren’t buying Mark Zuckerberg’s AI future.
💸 DeepSeek's AI models are about to cost four times more.
🤦 Report supporting Australia’s teen social media ban appears to contain AI hallucinations, Senate hears.
🤝 SpaceX has officially acquired AI coding startup Cursor.
🩺 Samsung unveils Health AI Models that read your body 24/7.
😱 Deepfake Anthony Albanese used in celebrity scams, duping Australians out of $7.4m, Asic warns.
🗣️ The first anti-AI protester to be jailed sent a message for OpenAI, Anthropic and Meta.
🤑 Stripe will reportedly acquire AI gateway startup OpenRouter for $7B+
🛠️ Trending Tools
For the Developers & Technical Architects: Grok 4.6 features a massive 500,000-token context window designed specifically for complex coding and multi-step agentic tasks. Available directly in Cursor and Grok Build, it self-verifies its output while navigating repository-scale codebases and deep technical research. It’s an ideal model for developers running long-horizon programming loops without losing context.
For the Operations Leads & Automation Teams: GrokBot deploys active AI agents operating inside dedicated, cloud-based virtual environments. These agents log into web tools just like a human operator, navigating interfaces to finish multi-step workflows before returning for final verification. It lets teams automate back-office tasks across platforms without needing to write or maintain custom APIs.
For the Hospitality & Restaurant Managers: MenuForma instantly converts any menu photo, PDF, or web page into an interactive QR-code ordering portal. Its centralized dashboard manages contactless payments, table ordering, allergen tags, and automatic translations into over 70 languages. It gives food vendors an easy way to modernize guest ordering, capture reviews, and track performance in one place. Best for: Restaurant owners and hospitality managers upgrading their physical or digital menu setups.
🤖 AI Workout Of The Day: Custom Weekly Workout Plan Request
A workout plan shouldn't just be a random collection of exercises; it must be a structured physical adaptation system.
Generic, one-size-fits-all routines often lead to quick plateaus, overuse injuries, or burnout because they ignore your unique biomechanics, recovery capacity, and real-world schedule. A high-impact training plan optimizes movement patterns, manages volume and intensity through progressive overload, and aligns directly with your available equipment and lifestyle constraints, delivering maximum athletic return on your time investment.
💡 Prompts To Try:
Act as an elite strength and conditioning coach, exercise physiologist, and sports nutritionist.
Your objective is to design a highly personalized, progressive weekly training and recovery blueprint tailored to my specific physical profile and operational constraints:
* Profile: [AGE] year-old [GENDER], [CURRENT WEIGHT & HEIGHT, e.g., 175 lbs, 5'10"]
* Primary Goal: [e.g., Hypertrophy/Fat Loss, Strength & Power, Cardiovascular Endurance, Recomp]
* Training Schedule: [NUMBER] days/week for [DURATION, e.g., 45-60 minutes] per session
* Available Equipment: [e.g., Full commercial gym, Dumbbells & Resistance Bands, Bodyweight only]
* Physical Limitations/Injuries: [e.g., Lower back tightness, minor left shoulder impingement, None]
* Current Fitness Level: [Beginner, Intermediate, Advanced]
Please generate a complete 7-Day Training & Recovery Protocol structured into the following 5 distinct operational sections:
1. PERIODIZATION & SPLIT ARCHITECTURE: Define the weekly split structure (e.g., Upper/Lower, Push/Pull/Legs, Full Body) and explain the physiological reasoning behind this routine for my goals and recovery capacity.
2. DAY-BY-DAY EXERCISE MATRIX
For every workout day, construct a clean Markdown table containing:
* Exercise Name & Target Muscle Groups
* Sets & Rep Ranges (e.g., 3 sets x 8-10 reps)
* Target Intensity/RPE (Rate of Perceived Exertion on a 1-10 scale)
* Rest Periods between sets (e.g., 90-120 seconds)
* Substitution/Option: 1 alternative exercise for each movement if equipment is occupied or causes joint discomfort.
3. WARM-UP & MOBILITY PROTOCOL: Outline a 5-minute dynamic warm-up sequence specifically targeting the joints and muscle groups used in that day's session to prevent injury and optimize movement quality.
4. PROGRESSIVE OVERLOAD & INTENSITY RULES: Detail 2 exact rules for how I should progress week-over-week (e.g., adding weight, increasing reps, or slowing tempo) when an exercise becomes too easy.
5. RECOVERY, DELOAD & METABOLIC ALIGNMENT:
* Active Recovery Days: Provide specific guidelines for rest days (e.g., Zone 2 cardio, mobility routines, step counts).
* Red Flag Warnings: List 3 signs of overtraining or systemic fatigue and explain when to implement a deload week.
TONE & EXECUTION GUIDELINES:
* Approach this with an encouraging, highly analytical, and scientifically backed tone.
* Use Markdown tables, bold text, and bullet points to ensure the workout split is easy to read on a phone while at the gym.Is 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.
