• The Automated
  • Posts
  • DeepSeek’s Latest AI Plays It Safer Than Ever?

DeepSeek’s Latest AI Plays It Safer Than Ever?

Plus, learn how to nail the perfect closing line using AI

In partnership with

Welcome, Automaters!

🚨 This is the Final Day Before the Price Doubles

Starting tomorrow, the price goes up.

This is your last chance to lock in $9.99/month — before it’s gone for good.

So Upgrade now and don’t miss out!

👉 More info here.

Here's what we have for you today

 😱 DeepSeek’s New AI Dodges Even More Sensitive Topics

So, China’s DeepSeek just dropped an upgraded version of its R1 model—say hello to R1-0528—and performance-wise? It’s killing it.

We’re talking near-OpenAI-o3 level scores on coding, math, and general knowledge.

But before we hand it a medal, here’s the twist:

While it’s getting brainier, it’s also getting way more tight-lipped about certain topics. Especially the ones the Chinese government would rather not talk about. Yeah… that kind of tight-lipped.

According to “xlr8harder”—a developer known for testing how AI models handle sensitive questions—DeepSeek’s latest brainchild is also its most censored model yet. Like, seriously sanitized.

And not in a “let’s keep things civil” kind of way—more like a “don’t talk about that or we both disappear” situation.

  • It straight-up refuses to engage with “contentious” topics—especially anything even slightly critical of the Chinese government.

For example? Ask about the Uyghur internment camps in Xinjiang, and the model either dodges the question completely or spits out the official government line.

  • Sure—sometimes, it gets brave for a second and mentions “human rights concerns”… but immediately retreats to the safe, state-approved narrative.

And no, this isn’t paranoia. China literally passed a 2023 law that bans AI models from generating content that could “damage national unity” or “undermine social harmony.”

That's why DeepSeek (and other Chinese AI companies) play it safe with heavy prompt filtering and fine-tuning to make sure their models don’t say anything “illegal.”

Even Chinese video-generating models like Magi-1 and Kling have been called out for doing digital gymnastics around historical events.

And it’s not just xlr8harder saying this. TechCrunch did some testing too—and yep, their quick experiments confirmed the model is locked down tighter than a middle school diary.

In Short:

DeepSeek’s R1-0528 is a genius—with boundaries.

It might outsmart half the models out there… but it also comes with a list of “do not speak about” topics longer than your Netflix watchlist.

Even Hugging Face’s CEO, Clément Delangue, is waving the caution flag. He warned that Western developers should think twice before building on top of Chinese models—because powerful AI with baked-in censorship?

Yeah, that gets awkward fast when you're dealing with open-source values and free speech.

But let’s toss this to you:

👉 As a developer, does the censorship bother you? Do you think it limits innovation or critical thinking? Would you still build with it—or is that a dealbreaker?

Before you answer that... maybe take a minute to really think about the trade-offs, and of course dive into the full details here.

Try Artisan’s All-in-one Outbound Sales Platform & AI BDR

Ava automates your entire outbound demand generation so you can get leads delivered to your inbox on autopilot. She operates within the Artisan platform, which consolidates every tool you need for outbound:

  • 300M+ High-Quality B2B Prospects, including E-Commerce and Local Business Leads

  • Automated Lead Enrichment With 10+ Data Sources

  • Full Email Deliverability Management

  • Multi-Channel Outreach Across Email & LinkedIn

  • Human-Level Personalization

🦾 Perplexity’s Latest Power Move.

You know how Perplexity’s been gunning for Google in the AI search race? Well, they just dropped a new weapon:

Perplexity Labs — a shiny new feature for Pro users ($20/month) that goes way beyond search.

According to reports, it’s basically like handing your laptop a personal assistant who actually knows what they’re doing.

You give it a task, and it runs with it for 10+ minutes straight — using AI to pull off some seriously grown-up stuff.

In short the new tool can:

  1. Research topics and summarize findings

  2. Generate full reports and documents

  3. Create spreadsheets with formulas

  4. Code and structure data

  5. Build interactive web apps (yes, actual apps)

  6. Make charts, visuals, and dashboards

  7. Organize everything it creates so you can download it later like a boss 

Even better? It’s already live on web, iOS, and Android — with desktop apps for Mac and Windows dropping soon.

Now here’s what makes this more than just a shiny new release:

Perplexity’s clearly making a hard pivot into AI-powered productivity.

They’re testing a new browser called Comet, acquired the professional network Read.vc, and they’re slowly building an AI ecosystem that’s less “search bar,” more “do-it-all digital brain.”

Oh, and yeah — they’re raising the stakes. Literally.

Rumor has it they’re in talks to raise up to $1 billion at a spicy $18 billion valuation. VCs are thirsty for returns, and Perplexity’s out here proving it’s more than just a Google clone.

But real talk — AI still fumbles a lot. So this won’t be flawless. Still, Perplexity Labs is definitely shooting its shot.

And if it actually works the way they say it will, it might just become a formidable competitor.

Here’s the full breakdown.

🎁 Referral Radar

Still haven’t picked up your gifts?

You’ve got this weekend to grab them—all it takes is 1–2 or 3 referrals and they’re yours!

🧱Around The AI Block

  • 🗣️ How to build a voice assistant with the Agents SDK.

  • 🤝 New York Times partners with Amazon for first AI licensing deal.

  • 👀 Gemini can now watch Google Drive videos for you.

  • 👩‍💻 Razer is making its AI tools available to all developers.

  • 🎉 Meta AI now has 1B monthly active users.

  • 🤖 Hugging Face unveils two new humanoid robots.

  • 🕵️‍♂️ Delaware attorney general reportedly hires a bank to evaluate OpenAI’s restructuring plan.

  • 🤑 Nvidia, AMD may soon start selling new AI chips in China to comply with US restrictions.

🤖 ChatGPT Prompt Of The Day: How to Close Like a Pro

You’ve delivered value, kept your audience engaged… Now what? Endings matter just as much as openings. A powerful closing line can inspire action, spark reflection, or leave a lasting impression long after the room clears or the video ends.

Whether it’s a webinar, presentation, or piece of content, your final words are your legacy moment. Use this prompt to craft a mic-drop closer your audience will carry with them long after the last slide or sentence

Here’s how to execute this prompt effectively:

  1. Clearly define the topic or message you want to leave your audience with.

  2. Choose the tone: Do you want to be motivational, thought-provoking, emotional, or bold?

  3. Think about the desired outcome — should your audience take action, feel empowered, or simply remember your key point?

  4. Use rhythm, repetition, or a callback to your opening for extra impact.

  5. Keep it short, punchy, and emotionally resonant — something that sticks.

Here’s an example:

Write a closing line that will leave my audience thinking, inspired, or ready to take action after hearing about [insert topic]. The tone should be [insert tone: motivational, heartfelt, bold, reflective, etc.]. Make it memorable — something that could be quoted, tweeted, or shared.

Here’s a sneak peek:

Upgrade now to see the complete prompt video. And yes do keep the great feedback coming — we’re in this together! 🚀

Is this your AI Workout of the Week (WoW)? Cast your vote!

Login or Subscribe to participate in polls.

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?

Login or Subscribe to participate in polls.

Your feedback means a lot to us and helps improve the quality of our newsletter.

🚀 Want your daily AI workout?

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....

Reply

or to participate.