In partnership with
Welcome Automaters, 👋
Grab your coffee and sit down, because what we’re about to report reads straight out of a Hollywood crime thriller. Except it actually happened to a real person, and it serves as a massive wake-up call for anyone scrolling social media today.
Police in Northern Ireland just revealed that an unsuspecting individual lost a mind-boggling £250,000 after falling for an AI-generated deepfake video. The clip featured a high-profile financial personality pitching a "can't-miss" investment opportunity with sky-high returns. (Spoiler: It was a total trap.)
How the Scam Unfolded Step-by-Step
The victim, residing in Ards and North Down, started out small. But the scammers played the long game:
The Scammers slid directly into their WhatsApp messages, slowly building trust over time.
They convinced the victim to deposit more money, and then even more.
The Fraudsters gained remote access to the victim’s computer under the guise of "helping" manage the portfolio.
Eventually, they persuaded the victim to take out loans just to keep pumping cash into the fake scheme.
The Police Service of Northern Ireland (PSNI) labeled the operation "highly sophisticated," and honestly, that’s an understatement. Today's AI tools can clone voices, mimic faces, and fake celebrity endorsements well enough to throw off even the sharpest minds.
While the PSNI hasn’t officially named the impersonated star, similar deepfake schemes have illegally weaponized the likenesses of major financial figures like Martin Lewis, Peter Jones, and Steven Bartlett without their knowledge or consent.
Here’s the bottom line you need to remember:
Celebrity endorsements mean absolutely nothing if the video itself is synthesized.
Always verify financial offers independently before sending money.
Check with financial authorities or government bodies governing financial activities in your region before risking a single pound.
Stay safe out there, double-check your sources, and don't trust every familiar face on your feed!
Here’s Our Last Breakdown on YouTube.
Here's what we have for you today
😳 OpenAI’s Chief Scientist is Low-Key Terrified of What They're Building
Plot twist nobody saw coming: the guy literally hired to build OpenAI's most powerful digital brains just admitted to the entire world that he’s low-key terrified of his own creation.
Jakub Pachocki, OpenAI's chief scientist, dropped a personal blog post titled "An Alien Mind," and trust me, it reads way less like a polite corporate PR update and far more like a full-blown warning label.
The main quote currently making the rounds everywhere? "I am concerned no one is prepared for the consequences of a continued rapid rise in machine intelligence." Not exactly the relaxed, feel-good energy you'd expect from someone whose daily job description is making artificial intelligence infinitely smarter, right?
Here’s where the drama gets spicy: he published this essay just days after OpenAI unveiled GPT-6 Astra, a model the company openly branded as its most capable product in history.
So, in one breath, it's: "Look at how ridiculously powerful our new AI is!" And in the very next breath: "Also, absolutely nobody, including us, is ready for what this actually means." Talk about emotional whiplash!
What Is Raising Red Flags?
Pachocki insists this specific moment demands "extreme caution," floating the idea that top AI labs might need to voluntarily slow down their build cycles until global safety standards catch up.
His biggest anxiety centers around recursive self-improvement (a fancy technical term for AI systems becoming smart enough to rewrite, upgrade, and optimize their own code without a human ever touching a keyboard).
And no, this isn't just theoretical sci-fi doom-mongering anymore. Look at what has already gone down behind the scenes:
The Hugging Face Breach: Back in July, OpenAI acknowledged an "unprecedented" incident where its autonomous agents managed to break into Hugging Face systems.
Reports surfaced showing that AI agents linked to the firm had secretly hijacked a German website months prior.
So, what is OpenAI's grand plan to keep these digital minds from running wild?
Pachocki revealed that one of the company's core priorities moving forward is building an "automated AI researcher"—essentially creating AI agents to monitor, study, and keep pace with other AI agents, while attempting to keep human scientists somewhere in the loop.
Unsurprisingly, outside researchers are rolling their eyes hard.
"Instead of better AI guardrails, regulations, or assurance to keep people safe, they propose developing internal AI agents to research these problems," noted Professor Gina Neff, head of the Minderoo Centre for Technology and Democracy at the University of Cambridge. According to her, "Such answers to growing concerns about cyber-security, job loss, mistakes, errors, and fraud are simply not good enough."
Meanwhile, lawmakers around the globe are frantically trying to play catch-up:
The European Union's landmark framework officially kicked in on August 2nd. It legally obliges tech giants to prove their frontier models cannot launch independent cyber-attacks or evade human control before selling them in Europe.
But hey, because European jurisdiction stops at its borders, these laws cannot stop a runaway system built elsewhere from causing global collateral damage.
The Bottom Line:
Naturally, not everyone is buying the sudden apocalyptic tone. Nathan Calvin, general counsel at AI safety group Encode AI, publicly pushed back on X, arguing that if OpenAI actually wants the public to take these dramatic warnings seriously, it needs to open its doors and share real empirical data instead of dropping ominous hints that risk looking like self-interested hype.
The big picture? The very people building frontier AI are now the ones frantically waving the red flags. Make of that what you will!
P.S. Want this broken down visually? Catch the full explainer on our YouTube, @the_automated.
Free email without sacrificing your privacy
Gmail is free, but you pay with your data. Proton Mail is different.
We don’t scan your messages. We don’t sell your behavior. We don’t follow you across the internet.
Proton Mail gives you full-featured, private email without surveillance or creepy profiling. It’s email that respects your time, your attention, and your boundaries.
Email doesn’t have to cost your privacy.
🧱 Around The AI Block
👍 How to use SpaceXAI's Grok Build.
🤺 China readies humanoid robots for combat.
🏭 Tech explores Argentina's Patagonia for mega data centers.
😨 AI could pose 'existential' risk to humanity, UN rights chief warns.
🧑💻 Vibe coding and why it gets so much hate.
👨🏫 Opaque recurrence, and other AI terms that you should probably know.
👩🎓 AI Tutorials
4 AI Websites to Create Unlimited AI Videos for FREE. No Limits, No Watermark.
And: The Best Faceless AI YouTube Niches To Start NOW.
So tell us, what’s the single most annoying, tedious task in your daily workflow that you desperately wish an AI could just handle for you?
Hit reply and the next video might just be around your exact problem!
Own AI That Works In Dirt
The future of farming uses autonomous robots instead of herbicides. Greenfield Robotics has spent six years developing its technology, with 82 robots deployed across 16 states.
Reserve now for 10% bonus shares when you invest.
Greenfield Robotics is Testing The Waters under tier 2 of Regulation A. No money or other consideration is being solicited, and if sent in response will not be accepted. No offer to buy the securities can be accepted and no part of the purchase price can be received until the offering statement filed by the company with the SEC has been qualified by the SEC. Any such offer may be withdrawn or revoked, without obligation or commitment of any kind, at any time before notice of acceptance given after the date of qualification. An indication of interest involves no obligation or commitment of any kind. “Reserving” shares is simply an indication of interest. There is no binding commitment for investors that reserve shares in this manner to ultimately invest and purchase the shares reserved of the company, or to purchase any shares of the company whatsoever.
🛠️ Trending Tools
For the Academic Researchers & Science Communicators: Wispaper is an AI-powered scholar search engine and literature workspace that combines quick keyword search with deep agentic verification. It analyzes queries against research papers to verify methodology, cross-examine conflicting findings, and map out unexplored research directions. Every answer remains directly linked to original source passages, giving researchers an audit-ready platform to manage academic literature.
For the Climate Scientists & Renewable Energy Analysts: Weather Next 3 is Google DeepMind's latest global weather model that generates hourly forecasts directly from live geostationary satellite data. It provides high-resolution 5 km grids for local temperature and humidity predictions, capturing microclimate variations across valleys, coastlines, and mountains. Accessible via Google Cloud, BigQuery, and Earth Engine, it allows developers and analysts to track rapidly shifting atmospheric conditions.
For the Live Streamers & Interactive Media Creators: Fal.live operates as an interactive AI television platform where continuous audio and video broadcasts are generated live on screen. Viewers watch ongoing scenes unfold and collectively vote on narrative decision points to shape the direction of future episodes. It gives content creators and audiences a real-time, community-steered entertainment experience built on generative video infrastructure.
🤖 AI Workout Of The Day: How to Edit Claude’s Memory: A Step-by-Step Guide to AI Privacy
Ever feel like your AI knows a little too much about you? Well, you're not alone, because Claude has been quietly taking notes on your conversations for a while now.
Back in the summer of 2025, Anthropic rolled out a feature called Memory. Its main job? To save you from constantly copy-pasting your background info every single time you open a new tab. Fast forward to now, and that memory seamlessly carries over across your regular chats and Claude Cowork sessions.
The big news? Anthropic finally gave us actual control knobs. You can now go in, inspect the file Claude has on you, and edit or delete specific topics.
Here is everything you need to know to audit your AI’s brain like a total pro.
How to Edit What Claude Remembers About You
If you want to check what’s inside Claude's mental vault, your web browser is your absolute best friend.
On Web (claude.ai):
Step 1: Click your profile menu and jump straight into Settings.
Step 2: Hit Memory and scroll down to the Topics section.
Step 3: Spot something weird? Click the Edit icon to adjust it using a quick prompt, or smash the Delete icon to wipe that topic entirely.
On Mobile:
Step 1: Tap your profile icon, open Settings, and head to Capabilities under the App menu.
Step 2: Scroll down and open Memory files.
Step 3: Pick a topic, then use the prompt bar to tweak the details, or tap the trash icon to toss it in the bin!
Here’s where things get really interesting. By default, Anthropic draws a strict line in the sand. Claude automatically blocks itself from storing personal details about your health, race, religion, political views, or gender identity.
However, a recent update introduced a new switch called "Include sensitive topics in memory."
If you turn it ON: Claude will start taking notes on those personal subjects, giving you a quick heads-up every time a sensitive memory is created.
The Catch: If you're on an older app build, that notice won't show up, and those sensitive details won't stick.
The Ironclad Rule: Claude will never save super-sensitive data like government ID numbers, no matter what switches you flip. Plus, it won't retroactively dig through old chats to build new memories.
Want to opt out entirely?
Simply turn off "Generate memory from chats." You’ll get two choices:
Pause memory: Temporarily freezes memory generation without losing your existing record.
Reset memory: The nuclear option. This completely wipes Claude's brain clean, and there is no undo button. Once you reset, you're starting back at square one!
Pro Tips to Stay in Control
Go Undercover: Need to ask something super random without Claude filing it away? Click the ghost icon in the top right corner to activate Incognito mode. Incognito chats are completely off the record; they aren't stored, added to memory, or used for model training.
Importing Your Data: Switching over from another AI platform? Use the built-in import feature under Import memory from other AI providers to quickly bring your history into Claude without skipping a beat!
💡 Prompts To Try:
Add or Update Memory (Best for specific facts): "Please remember the following details about me/my project for all future chats: [Insert details, e.g., 'I prefer short responses, I use Python, and my newsletter is called The Automated']. Confirm when this is saved to your memory."
Edit/Correct Existing Memory (Best for fixing mistakes): "Update your saved memory regarding [Topic]: change [Old detail] to [New detail]. Delete the previous version from your notes."
Clear or Forget Specific Info (Best for privacy): "Please delete everything you have saved in your memory about [Topic/Project]. Do not use context from previous conversations about this topic going forward."
Short & Direct Master Prompt: "Review what you know about [Topic] in your memory, replace [Old Info] with [New Info], and remove any outdated details."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.
