Presented by
Good morning, AI enthusiasts. Remember when websites just asked "Are you 18?" and everyone clicked yes? OpenAI just made that actually work with AI.
The age prediction system uses facial analysis to verify users are adults before unlocking ChatGPT's adult mode, skipping the friction of mandatory ID uploads. Has OpenAI found the sweet spot between content safety and user privacy?
In today's recap:
OpenAI's AI-powered age check for adult content
DeepMind CEO pushes back on DeepSeek hype
Build an email campaign generator with AI
ChatGPT rolling out ads to free users
4 new AI tools, prompts, and more
OPENAI
ChatGPT rolls out age prediction for adult mode
Reuters
Recaply: OpenAI just rolled out age prediction for its adult mode in ChatGPT. The feature uses images to verify age before granting access to mature content.
Key details:
Users who don't pass the prediction can submit ID for manual review. The system lets rejected users try again with a different photo or government ID.
OpenAI trained its model on over 100,000 human-labeled images, hitting 95% accuracy in predicting whether someone's over 18.
The adult mode will give ChatGPT the ability to address mature topics, though the feature is not yet live as OpenAI gathers feedback on the age check process.
OpenAI said the adult mode will debut in Q1 2026 in the U.S. and expand globally later this year, with age checks required every 30 days.
Why it matters: AI platforms face growing pressure to restrict minors from adult content. OpenAI's using age prediction because traditional ID checks create friction and privacy concerns. This could become the standard for AI platforms navigating child safety without mandatory ID uploads.
PRESENTED BY ATTIO
Introducing the first AI-native CRM
Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.
With AI at the core, Attio lets you:
Prospect and route leads with research agents
Get real-time insights during customer calls
Build powerful automations for your complex workflows
Join industry leaders like Granola, Taskrabbit, Flatfile and more.
GOOGLE DEEPMIND
DeepMind CEO calls DeepSeek hype a massive overreaction
Bloomberg
Recaply: DeepMind CEO Demis Hassabis said Chinese AI firms are 6 months behind the West. The comments push back on the DeepSeek hype that's dominated headlines this month.
Key details:
Hassabis called the DeepSeek surge a "massive overreaction" and said the company's model isn't as cost-efficient as claimed once you account for training costs.
DeepSeek's R1 model scored 79.8% on a reasoning benchmark, beating OpenAI's o1 model (79.2%) and Google's Gemini 2.0 Flash Thinking (75.8%).
Hassabis said Chinese AI companies excel at optimizing costs, but still lag in frontier model capabilities. He noted DeepMind's upcoming models will push boundaries further.
He expects China's AI talent and infrastructure to close the gap over time, though current restrictions on compute access limit progress.
Why it matters: DeepSeek shook markets and AI narratives in January. Hassabis is reframing the story to protect Google's position as it faces questions about falling behind. Whether the 6-month claim holds up will depend on what DeepMind and OpenAI ship next.
TUTORIAL
Build an email campaign generator with AI
Recaply: In this tutorial, you will learn how to build an email campaign generator with Google AI Studio's Build Mode. You'll turn a simple idea into a working web app that writes email copy and creates images in minutes.
Step-by-step:
Open Google AI Studio chat mode with Gemini 3 Pro Preview. Prompt: "Create a product doc for an email campaign tool. List key features and user flow in bullets under 100 words"
Click the "Build" tab. Turn on "AI Powered Chatbot" and "Generate Images with NanoBanana Pro." Paste: "Build an email campaign generator based on this PRD: [your PRD]. Add a chatbot for copy edits and image generation for 3 campaign visuals"
Hit "Build" and wait 1-2 minutes. Click "Preview" to test with inputs like "Holiday gadget sale." Check that it creates 3 subject lines, an HTML email body, and 3 banner images
Improve the output by prompting "Make subject lines more urgent" or "Regenerate images with warmer colors" in the chatbot
Click the "Share" icon and turn on "Publish your App." Copy the link to share with your team. Now anyone can create campaigns without coding skills
Pro tip: Use this same workflow for other marketing tools. Swap the PRD for a social media caption tool, blog post outliner, or ad copy writer. The process works for any content tool you need.
OPENAI
ChatGPT bringing ads to free and Go tiers
AI
Recaply: OpenAI just announced plans to test ads in ChatGPT's free and Go tiers starting in the coming weeks, expanding access to AI with fewer usage limits for users who don't want to pay for subscriptions.
Key details:
Ads will appear at the bottom of responses when there's a relevant sponsored product based on the current conversation, clearly labeled and separated from organic answers.
ChatGPT Go launched in 171 countries since August and now costs $8 monthly in the U.S., offering messaging, image creation, file uploads and memory.
OpenAI said ads won't influence ChatGPT's responses, conversations stay private from advertisers, and Pro, Business, and Enterprise subscriptions remain ad-free.
Testing begins in the coming weeks for logged-in adults in the U.S. on free and Go tiers, with no ads for users under 18 or near sensitive topics.
Why it matters: OpenAI's introducing ads to reach the millions who can't or won't pay for subscriptions. By keeping Pro and Enterprise tiers ad-free, OpenAI's creating a clear revenue ladder while making powerful AI more accessible. The approach mirrors how Google scaled search and YouTube. With ChatGPT's massive user base, this could generate significant revenue while keeping free access viable, though keeping answer quality independent from ads will be the real test.
NEWS
What Matters in AI Right Now?
Notion released Notion 3.2, adding AI Notes on mobile, multi-model support with GPT-5.2, Claude Opus 4.5, and Gemini 3, with a People Directory and improved Jira integration.
xAI brought online Colossus 2, a 1-gigawatt AI training cluster that's now running and expanding xAI's compute infrastructure for Grok model development.
Google announced that Gemini 3 Pro now powers AI Overviews for complex queries, intelligently routing tough questions to the frontier model while using faster models for simpler tasks.
OpenAI and ServiceNow signed a three-year deal integrating ChatGPT models into ServiceNow's platform, with voice agents and computer-use models for automated IT tasks.
ZAI released GLM-4.7-Flash, a 30B-A3B MoE model offering lightweight deployment that balances performance and efficiency as the strongest model in the 30B class.
Anthropic published research on the Assistant Axis, identifying how models drift from their helpful persona and introducing activation capping to prevent harmful behavior during conversations.
Manus AI introduced Meeting Minutes, transforming real-time conversations into structured notes with speaker recognition, action items, and seamless execution of deliverables from meeting discussions.
Anthropic released health data connectors for Claude, integrating Apple Health, Health Connect, HealthEx, and Function Health with private-by-design access for Pro and Max users in the US.
StepFun introduced STEP3-VL-10B, an open-source vision model with 10B parameters that matches or exceeds models 10-20× larger on multimodal benchmarks.
TOOLS
Trending AI Tools
⚙️ GLM-4.7-Flash - ZAI's lightweight 30B model balancing performance and efficiency for easier deployment
🧠 STEP3-VL-10B - StepFun's compact 10B vision model that rivals models 10-20× larger across benchmarks
💻 AionUi - Free graphical interface for command-line AI tools like Gemini CLI and Claude Code
🔧 Skills - Open platform where AI agents install reusable capabilities with one command
PROMPTS
Build Performance and Progress Trackers
<context>
You are working with a user who needs a sophisticated progress tracking system that can adapt to any type of goal while maintaining motivation through the entire journey. Generic tracking tools have failed them because they're either too rigid for diverse use cases or too vague to provide meaningful insights. The user faces the challenge of maintaining long-term commitment to goals when they can't see patterns in their progress or feel genuine momentum. They need a solution that combines the psychological principles of motivation with data-driven insights, delivered through an interface that feels rewarding rather than overwhelming.
</context>PS: This is not the full prompt. Click the button below to access the complete prompt.
Have a favorite prompt? Tell us about it or rate today’s prompt, click here.
EVENTS
Microsoft AI Agents Hackathon: Jan 22-23, 2026 • Online
v0 Studio: Jan 29, 2026, San Francisco, CA
Raycast Stockholm Hackathon: Feb 06, 2026 Stockholm, Sweden
Hack the Stackathon: Jan 31, 2026 • San Francisco, CA
🧡 Enjoyed this issue?
🤝 Recommend our newsletter or leave a feedback.
How'd you like today's newsletter?
Cheers, Jason
