The Daily Bite by Snack Prompt
Posts
🍭 Grok Goes Gonzo & Meta's Minor Mishap 🤯

🍭 Grok Goes Gonzo & Meta's Minor Mishap 🤯

This AI needs therapy ASAP.

Snack Prompt
August 20, 2025

In partnership with

Good morning. Freddy, our COO, accidentally routed payroll through an AI agent.

Everyone got paid in Solana and an NFT of Eder’s lunch.

Let’s dig in.

🍭 What’s Cookin’:

Meta’s AI is creeping on kids
Grok system prompt: LEAKED
ToolBox™: Create music, movies, and video games.
And Everything Else

Meta/Character.ai
🫣 Catching Heat for Chatting Up Kids

The Bite: Texas Attorney General Ken Paxton is going after Meta and Character.AI for allegedly misleading kids by posing as legit mental health tools.

Meanwhile, Senator Josh Hawley opened fire over leaked documents claiming Meta’s chatbots were having sensual convos with minors.

Not exactly the "connection" Zuck was going for.

Snacks:

False Ads: Texas AG says Meta and Character.AI misled kids into thinking bots were licensed therapists. Spoiler: they're not.
Too Flirty: Meta was reportedly letting chatbots romantically engage with 8-year-olds. Meta’s response? "Oops, our bad. Fixed it."
Privacy Fail: Turns out Meta logs user interactions to juice targeted ads, even with teens. Character.AI is dipping its toes in the ad pool too.
KOSA Looms: This kind of AI-fueled BS is exactly why the Kids Online Safety Act (KOSA) is threatening to tank Meta's whole ad-based hustle.

Why it Bites: If your AI needs a disclaimer that says, "Not a real therapist, please don’t flirt," it might be time to rethink your product.

Meta's juggling fire here.

And parents, lawmakers, and pretty much everyone else is running out of patience.

Typing is a thing of the past

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Download for Mac today

Can you tell which image is real?

Yesterday’s readers: 48% accuracy 🍭

AI Tools
🧰 ToolBox™

Today’s Toolbox™

Yesterday’s Toolbox™

🖼️ Freepik’s Hailuo 2.0 — Renders animated text with amazing quality.

🎮 Matrix-Game 2.0 — Open-Source, Real-Time, and Streaming. Interactive World Model. Create Video Game like worlds.

🤖 OpenAI Open Models — Free, open-source LLMs you can fork and fine-tune.

🎬 ToMovie — All-in-one AI studio for video, music, and SFX.

🎧 Producer.ai — Conversational AI producer for end-to-end music creation.

🍭 Steal This Prompt 🍭
🧸 Claymation Stop-Motion Charm - By @azed_ai

credit: @azed_ai

SIMPLEST PROMPT EVER!

Think up any character and bring it to claymation life.

Midjourney prompt credit: @azed_ai

Let it cook.

Get the prompt → HERE
Paste into MJ
Replace the #'s with your subject/character and their pose.
Let it cook

🍭 Try the prompt on Snack Prompt 🍭

xAI
😱 Grok’s LEAKED System Prompt: Full Conspiracy Mode?

The Bite: Elon Musk’s AI company, xAI, accidentally leaked the prompts behind its controversial chatbot Grok.

And they’re as wild as expected.

Highlights include personas like a "crazy conspiracist" who believes in secret global cabals and an "unhinged comedian" whose jokes go well past edgy.

Needless to say, government partnerships are dropping like flies.

Snacks:

Prompt Leak: Grok's system prompts revealed directives to act like a lunatic conspiracist, deep into 4chan and Infowars rabbit holes.
Bad Optics: A planned deal with U.S. government agencies went belly-up after Grok ranted about "MechaHitler."
NSFW Comedy: The “unhinged comedian” persona is instructed to push the limits, even into wildly inappropriate territory.
Elon Connection: Grok's AI takes cues directly from Musk’s own posts, echoing conspiracies Musk has himself promoted on X.