104: Today in AI: ChatGPT Ads Hit Paid Users, DeepSeek Rivals GPT-4 for Pennies, Claude Limits Anger

2025-12-05

Join 40,000 readers and get the full notes and summary: https://newsletter.aiwithkyle.com/subscribe Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/@iamkylebalmer?sub_confirmation=1 Want to know which AI business to start? Check out Freedom OS: https://aiwithkyle.com/join-freedom-os This video is from the ‘AI with Kyle News and Updates Live Stream’ First aired: 4 December 2025 Summary: OpenAI found themselves in a bit of a pickle this week when paid Cha...

Join 40,000 readers and get the full notes and summary: https://newsletter.aiwithkyle.com/subscribe

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/@iamkylebalmer?sub_confirmation=1

Want to know which AI business to start? Check out Freedom OS: https://aiwithkyle.com/join-freedom-os

This video is from the ‘AI with Kyle News and Updates Live Stream’

First aired: 4 December 2025

Summary: OpenAI found themselves in a bit of a pickle this week when paid ChatGPT users started seeing what looked suspiciously like ads in their chats. Someone was asking about Windows BitLocker and suddenly got served a Target shopping ad at the bottom of his conversation—despite paying $20 a month for Plus. OpenAI's Chief Research Officer jumped in to clarify these aren't actually ads, they're "apps" from pilot partners like Target and Peloton, meant to augment the user experience. But here's the thing: if it looks like an ad, smells like an ad, and interrupts your conversation to suggest you go shopping... it's probably an ad, regardless of what corporate terminology you dress it up in.

Meanwhile, China's DeepSeek dropped their 3.2 model that's performing at GPT-4 levels whilst costing about 1/20th the price—which is genuinely impressive and a bit concerning for Western AI companies. And in Claude news, Opus 4.5 has finally rolled out to Pro users in Claude Code, though the rate limits are so restrictive that one prompt might be your lot for the week. It's brilliant when you can use it, but forcing paying customers to switch to competitors mid-task because you've run out of capacity is still a massive product design flaw, no matter how good the model is.

Watch full edited live stream: https://youtu.be/rYymJ-ALB3Q

—— More Useful Resources ——

My other daily live show - 30-day building a business in public challenge: https://www.youtube.com/live/Z0Z_1Eg3HUk (day one)

Ultimate AI Learning Resource List: https://www.aiwithkyle.com/resources/ai-learning-guide

Free AI Entrepreneurship Playbooks: https://aiwithkyle.com/catalog

Free Quiz: What AI business should you start? https://aiwithkyle.com/tools

Free AI Training Business Planner: https://aiwithkyle.com/tools/workshop-builder

Free AI Industry Translator Tool: https://aiwithkyle.com/tools/industry-ai-translator

Free 10-Week ‘Vibe Coding’ AI Summer Camp: https://aiwithkyle.com/courses/10-week-ai-summer-camp

—— Time Stamps ——

00:00 Cold Open: Google's Ad-Free Strategy

00:15 Welcome to AI with Kyle