• Duo Discover
  • Posts
  • OpenAI's new image generator hits different...

OpenAI's new image generator hits different...

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

Dear Tech Enthusiasts,

While Google's Gemini 2.5 Pro and several impressive Chinese models have quietly revolutionized AI, everyone's attention has been captivated by OpenAI's new GPT-4.0 image generator—a tool that's transforming the internet into what some call a "GBI anime cartoon nightmare."

The Anime Takeover

The famous animation director Hayao Miyazaki once warned: "I am utterly disgusted. If you really want to make creepy stuff, you can go ahead and do it. I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself."

His prophetic words now seem eerily relevant as social media platforms become flooded with AI-generated anime-style imagery. Even your favorite memes haven't been spared from this transformation.

What Makes GPT-4.0's Image Generator Different?

Despite initial skepticism following the underwhelming releases of Sora and GPT-4.5, OpenAI has delivered something truly impressive:

  • Near-perfect text rendering: Create infographics and marketing materials that previously required dedicated design software

  • Transparency handling: Unlike most generators, it can handle layers and transparency effects

  • Character continuity: The ability to maintain consistent character features across multiple images

  • Style transformation: Apply specific art styles to existing images

The Technical Innovation

Unlike diffusion-based models (Stable Diffusion, Midjourney) that generate entire images simultaneously, GPT-4.0 uses an autoregressive approach—generating the image pixel by pixel, from left to right and top to bottom. This approach appears to deliver more coherent results.

The Watermark Controversy

These seemingly seamless images actually contain invisible watermarks through the Coalition for Content Provenance and Authenticity (C2PA). Upload any GPT-4.0 image to the C2PA tool, and you'll see it was generated by OpenAI, along with a history of modifications.

Camera manufacturers and software developers like Adobe are implementing this technology to track every change to digital assets—ostensibly to combat misinformation, though critics argue it comes at the cost of privacy and creative freedom.

The Philosophical Question

As platforms like YouTube and Steam now require disclosure of AI-generated assets, an interesting philosophical question emerges, dubbed "Slop's Razor" by internet commentators:

"Can you tell it's AI-generated by looking at it? If the answer is no, then it's indistinguishable from human work, thus no disclosure is needed. But if you answer yes, then it's visibly 'slop' and again, no AI disclosure is needed."

Meanwhile in the LLM Space...

While OpenAI's image generator dominates headlines, other significant developments include:

  • Google's Gemini 2.5 Pro: Reportedly matches Claude 3.7 for programming capabilities with a larger context window and better reasoning than OpenAI's models—available for free

  • Chinese AI acceleration: DeepSeek 3.1, Alibaba's Qwen 2.5 Omni (multimodal with "thinker-talker" architecture), Tencent's T1, and ByteDance's open-source Dapo reinforcement learning system

We're living in what some call a "vibe coder's paradise" where anyone can generate more code than they could possibly need—meaning real programmers might soon be more focused on reviewing and refactoring AI-generated code.

Are we one step closer to the singularity? What are your thoughts on these developments?

What did you think of this week's issue?

We take your feedback seriously.

Login or Subscribe to participate in polls.