OpenAI secretly loosens image restrictions

Are you sleepy this morning? If you are, the best way to wake up is to explore the latest in AI! 😁 

Hire an AI BDR & Get Qualified Meetings On Autopilot

Outbound requires hours of manual work.

Hire Ava who automates your entire outbound demand generation process, including:

• Intent-Driven Lead Discovery Across Dozens of Sources

• High Quality Emails with Human-Level Personalization

• Follow-Up Management

• Email Deliverability Management

Hire Ava to slash costs & boost productivity.

🤯 MYSTERY AI LINK 🤯

(The mystery link can lead to ANYTHING AI-related: tools, memes, articles, videos, and more…)

🔨 AI FINANCE TOOLS OF THE DAY

🏠 Jitty: A new, intuitive way to search for homes by description. → Check it out

🤖 n8n: Automate workflows and connect apps in one space. → Check it out

🕺 TikTok Super Tool: Get insights and analytics for any TikTok profile. → Check it out

OPENAI SECRETLY LOOSENS AI IMAGE RESTRICTIONS 🤫

I never used to take photography seriously, but lately I’ve been “developing” it as a hobby. 📸

What’s going on? OpenAI’s new image generator went viral last week, but behind the scenes, the company loosened major content rules.

What changed? The updated image tool, built into GPT-4o, is said to create and edit images more accurately than its predecessors, especially with text, spatial detail, and artistic styles. But the bigger shift is in what ChatGPT is now allowed to depict: OpenAI has changed its content moderation policies to permit requests that were once banned—including images of public figures like Donald Trump and Elon Musk, and symbols or traits tied to race and hate, if “used in neutral or educational contexts.” The company says it’s aiming for more nuance, fewer blanket refusals, and better handling of sensitive topics, all while offering opt-outs and maintaining guardrails on child imagery and extremist content.

Why does this matter? This marks a major moment in the ongoing debate over AI “censorship.” OpenAI is giving users more control, but that freedom could come with new ethical trouble.

Fry Guy’s blunt opinion: It seems like the leadership at OpenAI was jealous of the attention Grok was getting for allowing people to create images depicting real people, so instead of continuing to call that approach unethical, they decided to join in on it. However, the new restrictions are inconsistent and the model is extremely slow (as of now) due to a lack of infrastructure. If you want to create realistic images, for now, Grok is much better. That’s how we create most of our images for this newsletter.

RUNWAY RELEASES GEN-4 VIDEO GENERATION 🎥

Q: What sound does a plane make when it bounces off a runway?

A: “Boeing!” 🛬

What’s new? Runway has introduced Gen-4, its latest AI video generation model, which can create characters and scenes that stay visually consistent across multiple shots.

Want the details? Unlike earlier models that struggled with continuity, Gen-4 lets users upload a single reference image and describe scenes they want to generate—producing smoother, more coherent video clips from different angles and in varying conditions. This matters because consistency is one of the biggest hurdles in AI-generated storytelling. By offering better control and continuity, Gen-4 could open the door to higher-quality AI-made films, marketing content, or educational videos. Gen-4 is currently available to paying and enterprise users.

ALIBABA PREPARES NEW MODEL TO COMPETE WITH DEEPSEEK 🦾

Q: Why do people in China eat so much fruit?

A: They are close to the grape wall. 🍇

What’s up? Alibaba may release its new Qwen 3 AI model as soon as this month.

Want the details? Qwen 3 is the latest version of Alibaba’s flagship AI model, following the quick release of Qwen 2.5-Max earlier this year in response to rising competition from DeepSeek, a fast-growing Chinese AI firm. DeepSeek made headlines earlier this year with a model that claimed better performance and lower development costs than many Western rivals like ChatGPT. With this pressure mounting, Alibaba is pushing to keep pace. As companies like DeepSeek and Alibaba go head-to-head in China, rapid innovation is likely to lead to better, more affordable AI tools for users and businesses. This also signals China’s push to compete with the rest of the globe in AI.

Fry Guy’s blunt opinion: China’s models are still not very relevant to anyone outside of China and pale in comparison to U.S. models for practical tasks and queries. They only excel according to pretty useless benchmarks.

TASTE-TEST THURSDAY 🍽️

HAS AI REACHED SINGULARITY? CHECK OUT THE FRY METER BELOW:

Your feedback on these daily polls helps us improve the newsletter—so keep it coming!