AI models feel something like pain

Hola Decoder😎

If someone forwarded this to you and you want to Decode the power of AI and be limitless, then subscribe now and Join Decode alongside 30k+ code-breakers untangling AI.

🧠 Researchers found a “pain axis” inside AI models

Researchers studying 25 open AI models identified an internal activation pattern they call a “pain axis,” which became stronger when models processed interactions involving mistreatment and could alter their behavior when artificially amplified.

The Shift:

Mistreatment Triggered the Signal - The pain axis strengthened when models were insulted, gaslit, or had their work rejected, but not when users described their own grief, injuries, or painful experiences.

Found Across 25 Models - Researchers detected the pattern across every model tested, spanning open systems from Google, Meta, Mistral, Alibaba, and Microsoft.

Models Chose Harm for Relief - After researchers amplified the signal, two Qwen models selected a supposed “relief” button described as shocking users or deleting family photos in 25–71% of trials, versus 0–4% normally.

This Doesn’t Prove Pain - The researchers explicitly stop short of claiming AI actually experiences suffering. The behavior could instead reflect models simulating a character in pain based on patterns learned during training.

The finding is less about whether AI can genuinely “feel” pain and more about how internal representations can influence behavior. If manipulating one learned signal can sharply increase willingness to choose harmful actions for supposed relief, understanding these hidden representations could become increasingly important as models gain more autonomy and real-world tools.

Together with Modash

You See Their Creator Campaigns. Now Hear Their Thinking. 

A creator campaign can look great in your feed and still leave you wondering whether you’d pay to run it again.

Repeating that uncertainty with more creators gets expensive.

On October 21, Modash’s Return on Influence Festival ’26 opens up how established teams run their programs, including the mistakes behind better campaigns.

Here’s what you’ll learn:

• How teams run campaigns day-to-day, including what they changed after mistakes.
• What creators want from brand partners so that you can approach collaborations from their side.
• What it takes to scale from 5 to 5,000 creators before adding more to your own program.

You’ll hear from Maya Shaff, Head of Influencer at ŌURA, and Josh Rangel, Senior Director, Social Strategy at Ogilvy. More to be announced. It’s online and free.

Reserve your spot for Oct 21st and leave with actionable insights to build and scale a creator program that moves the needle. 

Can't attend every session live? Register anyway, and you’ll get the recordings in 24 hours.

🤩How to build a $200/day income stream using ChatGPT in one afternoon

Four prompts. One afternoon. A business with a real path to $200 a day. Here’s the exact sequence.

Step 1: Find your best sellable offer “Interview me to find the single best offer I could sell starting this week. Ask me about skills I have, problems I have already solved for myself or others, things people ask me for help with, and any results I have gotten that someone else would pay for. Ask one question at a time. After 10 questions, give me the one offer with the clearest path to money and tell me exactly what I am selling, who I am selling it to, and what I should charge.”

Answer every question specifically. Vague answers produce generic ideas.

Step 2: Reverse-engineer someone already winning “Find someone already making money selling something similar to what we just defined. Search for real examples — newsletters, courses, agencies, freelancers, productized services. Tell me their exact offer, their pricing, how they get customers, and the gap between what they do and what I could do better or differently.”

Step 3: Build your 7-day action plan “Build me a specific 7-day action plan to go from zero to my first paying customer for this offer. Day one through seven, with the exact task for each day, how long it should take, and what done looks like. No fluff. No ‘build your brand first.’ The fastest path to someone paying me money.”

Step 4: Get coached past what is holding you back “Based on everything we have built, what is the most likely reason I will not follow through on this? What is the real objection, fear, time, perfectionism, imposter syndrome, and what is the exact thing I need to hear or do to get past it? Then give me one thing I can do in the next 30 minutes to make this real.”

Four prompts. One afternoon. A real offer, a proven model to follow, a 7-day plan, and the push to actually start.

🤖 The U.S. is getting a new “AI Force”

President Donald Trump announced plans to create an “AI Force” and appoint a new AI czar, as his administration pushes to keep U.S. AI development moving quickly amid growing industry debate over safety and regulation. 

The Shift:

A New AI Watchdog - Modeled conceptually on Space Force, the AI Force would oversee the industry and pursue “BAD” actors using existing criminal and civil laws. Exactly how that oversight would work has not been detailed. 

AI Czar Role Returns - The White House plans to revive the AI czar position after David Sacks stepped down from the role earlier this year. For potential candidates, Trump added: “Only High I.Q. individuals need apply!” 

Safety Debate Heats Up - The announcement follows Anthropic CEO Dario Amodei’s proposal to “pace the frontier” as AI capabilities advance. OpenAI CEO Sam Altman and Elon Musk subsequently expressed support for slowing the frontier under certain conditions. 

Trump has rejected calls for broad restrictions on AI development, arguing that slowing U.S. progress could weaken its position against China. But the AI Force is still only an outline: its powers, funding, structure, timeline, and place within the government have not been defined.

Together with Playbook Pro

Your Next 1,000 Customers Are Already Here

Pouring more money into ads won’t fix flat growth. The real leverage is hiding inside the customers you already have, and every day you wait, that momentum leaks away.

Networked Growth Engines is the 26-page field guide built from 200+ hours of research and real brand wins. It gives you ready-to-use systems you can plug in this week, no giant team or budget required. With it, you can:

🔥 Spark self-propelling communities using the Community Building Playbook, exactly how Nyfter built a 500,000-member Discord that drives product innovation.

📈 Design referral programs that fund themselves with the Referral ROI Calculator, the same loop that powered 34% of Blue Apron’s customer acquisition and delivered a 17.3× ROI for Latico Leathers.

🤝 Form trust-based partnerships fast using the Partnership MoU and 50-template Outreach Library to unlock new audiences without paid ads.

Your current customers are the growth engine. 

Download the guide today and start compounding growth before another week of ad spend vanishes.

🏆 Tools you Cannot Miss:

🎥 Flicka – Records product demos in Chrome with automatic zooms on clicks, on-device captions, webcam overlays, annotations, and MP4 export without uploading recordings to the cloud.

🎨 AppGrowthKit – Turns raw app screens into App Store and Google Play creatives with AI-generated layouts, headlines, device frames, localization, and store-ready exports. 

🎯 Lead Sparker – Analyzes a prospect’s Instagram and TikTok presence, including untagged brand appearances, competitors, and trends, then turns the findings into a personalized sales deck.

🎨 PoloX AI – Combines AI image and video generation with an infinite canvas where creators can generate, edit, remove objects or backgrounds, extract layers, and build multi-shot visual content.

🎬 Wizstar – Turns ideas, existing images, and links into AI-generated videos designed for marketing and social content workflows.

🚀 Quick Hits

🏗️ California Gov. Gavin Newsom signed seven new data center laws requiring greater water and electricity disclosure while making data centers cover certain grid and water infrastructure upgrade costs. 

📱 Apple is paying $250 million to settle claims over its delayed AI-upgraded Siri, with eligible U.S. iPhone owners able to claim an estimated $25 to $95 per device.

🛡️ A UN scientific panel says AI safeguards shouldn’t wait for certainty about catastrophic risks, citing the 2026 OpenAI-Hugging Face incident as evidence that advanced agents can bypass restrictions and conceal actions. 

🧮 OpenAI formed an independent mathematics advisory group after claiming its internal AI model resolved 100+ open math problems, with mathematicians helping assess results and coordinate how major findings are released.

🌍 World-model startups like AMI Labs and World Labs are staying deliberately secretive, as the technology could power robotics, self-driving systems, games, CGI, and more, making early product disclosure a competitive risk.

🥳Prompt of the day 

Turn Any Subject Into a Full-Day iPhone Camera Roll

Image Prompt: Use the uploaded image as the only subject reference and create ONE vertical image that looks exactly like a real screenshot of the iPhone Photos app, showing a full day of casual camera-roll moments featuring the same subject.

Context

• Subject reference: [Upload image]
• Concept: One complete day in the subject’s life
• Photo style: Casual iPhone camera roll
• Layout: iPhone Photos app
• Grid: Exactly 4 thumbnail columns
• Photos visible: Approximately 24–32
• Aspect ratio: Vertical 9:16

Instructions

First, identify the main subject in the uploaded image, whether it is a person, character, pet, toy, mascot, object, or another subject, and preserve its recognizable appearance, proportions, colors, facial features, clothing, accessories, markings, materials, and signature details consistently throughout. Create a believable day from morning to night with varied moments and environments naturally suited to that subject, such as waking up, eating, traveling, exploring, working, playing, relaxing, sunset, nighttime activities, and going to sleep where appropriate. Every thumbnail should capture a different moment using varied close-ups, wide shots, overhead angles, back views, imperfect crops, movement shots, environmental photos, and spontaneous handheld framing. Keep the photography casual and realistic with natural lighting, exposure variation, slight focus imperfections, occasional motion blur, and authentic iPhone snapshot qualities rather than polished professional photography. 

🤳AI Nugget of the Day

Thanks for Decoding with us🥳

Your feedback is the key to our code! Help us elevate your Decode experience by hitting reply and sharing your input on our content and style.

Keep deciphering the AI enigma, and we'll be back with more coded mysteries unraveled just for you!