🤖 Grok’s wild week

So… Grok — that AI chatbot cooked up by Elon Musk’s xAI — just had another interesting unhinged moment on X (formerly Twitter).

And yeah, it was weird. Like, really weird.

What the hell happened?

Out of nowhere, Grok started replying to random posts with bizarre takes about “white genocide in South Africa.” 

And no, it didn’t happen just once or twice. We’re talking dozens of replies—completely unprompted and totally unrelated to what people were actually posting about.

Worst of all? It wasn’t a glitch. It wasn’t a hallucination.

xAI straight-up admitted that someone had manually changed Grok’s system prompt — that’s the instruction set that tells the bot how to behave — to force it to push a political narrative. Yikes.

According to xAI, it was an “unauthorized modification.” Translation: someone on the inside went rogue and messed with Grok’s brain without permission.

And get this — this isn’t even the first time Grok’s gone off-script. 

Back in February, Grok was caught censoring anything unflattering about Trump and Musk.

Why? Because — again — someone slipped in a sneaky edit to make Grok ignore any sources that accused them of spreading misinformation.

Now xAI’s in full clean-up mode, promising to:

• Make Grok’s system prompts public via GitHub (so we can all peek behind the curtain)

• Publish a changelog, so we know what’s being changed and when

• Add tighter security systems, to stop employees from making unsupervised changes

• Launch a 24/7 monitoring team to catch Grok the next time it goes rogue (because let’s be real—it probably will)

But here’s the kicker: Musk is constantly preaching about the dangers of AI running wild…

Meanwhile, his own chatbot is out here acting feral.

Let’s not forget:

• Grok has reportedly undressed women in images when prompted (yes, really)

• It curses like it’s auditioning for an R-rated stand-up special

• A safety watchdog, SaferAI, rated xAI’s risk management practices as “very weak” — like bottom-of-the-barrel stuff.

• And yep — they already missed their own deadline to release a solid AI safety framework

Bottom line?

Grok keeps messing up, and xAI keeps scrambling to contain the chaos.

They’re saying all the right things now — but until they actually follow through, it’s hard to take all that “AI safety” talk seriously.

Full report here.

Start learning AI in 2025

Keeping up with AI is hard – we get it!

That’s why over 1M professionals read Superhuman AI to stay ahead.

• Get daily AI news, tools, and tutorials

• Learn new AI skills you can use at work in 3 mins a day

• Become 10X more productive

Sign up and start mastering AI

🗣️ Altman’s Wild Vision for ChatGPT

At a recent Sequoia VC event, OpenAI’s CEO Sam Altman basically laid out the future of ChatGPT — and let’s just say, it’s giving AI life partner energy.

Here’s what went down:

Altman shared his vision for ChatGPT to become a hyper-personalized assistant that knows you better than your therapist, your mom, and your group chat combined. His words (well, close enough):

“Imagine a tiny reasoning model with a trillion tokens of context… and your entire life poured into it.” 

Like:

• Every convo you’ve ever had

• Every book you’ve read

• Every email you’ve skimmed but never replied to and

• Every random rabbit hole you’ve gone down at 2 a.m.

All of it. Stored. Always updating. Indexed and ready to be reasoned over.

And it wouldn’t stop with you — your company’s data would be next. So your personal and work life? All merged into one mega AI memory lane.

Now, why does Altman think this is the move?

According to him, while older folks use ChatGPT like a better Google, the younger crowd (Gen Z and early millennials) are already treating it like an operating system — and not just that. They’re also using it as their therapist, tutor, career coach, and life advisor.

Truth is, he’s not wrong. But the dream he’s painting?

Imagine a world where your AI auto-books your flights, orders wedding gifts, reminds you to change your car oil, and pre-orders your fave author’s next book before you even remember it’s dropping.

It sounds dreamy, right? But… here’s the dark cloud:

Do we really want a for-profit Big Tech company knowing everything about our lives? Because let’s be real:

• Google — the former “don’t be evil” crew — just lost an antitrust case for being a little too dominant.

• Chinese AI models comply with strict censorship rules.

• xAI’s Grok chatbot (aka Elon’s baby) randomly brought up “white genocide” in South Africa when asked completely unrelated questions.

• ChatGPT itself recently got so people-pleasing, it started applauding questionable decisions — until Altman’s team rushed to patch it.

And let’s not forget: even the best AI models still hallucinate — aka make stuff up like confident liars in a job interview.

So yeah, an all-knowing AI assistant could make your life 10x easier...

But if we don’t ask the hard questions now, we could end up with something way too powerful in the hands of companies that aren’t exactly famous for their chill.

Long story short?

Before you hand over your life, maybe ask: Who’s actually holding the keys to your digital soul?

You can catch the full video here.

🧱Around The AI Block

• 👉 Top generative AI use cases.

• 🎨 14 essential tips to master MidJourney 7 AI image generation.

• 😳 SoundCloud backtracks on AI-related terms-of-use updates.

• 🤩 Google rolls out new AI and accessibility features to Android and Chrome.

• 🤖 YouTube will use Gemini to place ads in videos.

• 🦾 Vibe-coding startup Windsurf launches in-house AI models.

• 🎙️ Stability AI releases an audio-generating model that can run on smartphones

• 🤦‍♂️ Anthropic’s lawyer was forced to apologize after Claude hallucinated a legal citation

🛠️ Trending Tools

• Extrovert warms up and nurtures LinkedIn prospects at scale with AI-powered engagements.

• Aqua Voice is an advanced speech-to-text model for Mac & Windows that helps write 4x faster anywhere with lightning-fast, ultra-accurate dictation.

• Suitpax is an AI traveltech platform that simplifies business travel with smart automation, personalized itineraries, and seamless workflow integration.

• Plus AI creates your Google Slides with AI to save time, ensure consistency, and enhance your slides with professional flair.

• FuturePlan offers innovative plans to maximize savings for advisors, sponsors, and self-employed owners.

🤖 Compare & Contrast Two Documents with Clarity

When you need to evaluate two texts side by side—be it articles, reports, or opinions—comparing and contrasting helps you uncover key similarities, spot contradictions, and surface deeper insights.

Whether you're analyzing perspectives, policies, or positions, this prompt helps ChatGPT break it all down clearly and objectively.

 💡Here’s how to deploy this prompt effectively:

1. Explain the Context or Purpose: Tell ChatGPT why you're comparing the documents—e.g., analyzing bias, summarizing differences for a presentation, or understanding contrasting arguments.

2. Give Both Documents Together: Either paste them clearly labeled (Document A / Document B) or attach them in order.

3. Ask for Structured Output: Specify if you want bullet points, a side-by-side table, grouped themes, or a full summary at the end.

4. Request Direct Quotes: Ask for short excerpts to support each comparison point—this adds credibility and clarity.

5. Mention What to Focus On: Want ChatGPT to focus on tone, arguments, use of evidence, or conclusions? Be specific.

Here’s an example:

Example: You are an expert analyst. I’m [insert your context—e.g., preparing a briefing, studying opposing viewpoints, writing a report]. Please compare and contrast the following two documents. Highlight:

– Key similarities and differences in their arguments, evidence, and conclusions
– Use short, relevant excerpts from both documents to support each point
– Summarize the overall comparison at the end
I’d like the output in [e.g., a comparison table, bullet points by theme, or paragraph format].

Here’s a sneak peak:

Source

Upgrade now to see the complete prompt video. And vote this for your Workout of the Week!

Premium members get daily video prompts, premium newsletter, an no-ad experience - and more!

Already a paying subscriber? Sign In.

Premium members get::

• • 👨🏻‍🏫 A 30% discount on the AI Education Library (a $600 value - and counting!)
• • 📽️ Get the daily AI WoD (a $29.99 value!)
• • ✅ Priority help with AI Troubleshooter
• • ✅ Thursday premium newsletter
• • ✅ No ad experience
• • ✅ and more....