The Rundown AI / Articles / AI / ChatGPT to go 18+

ChatGPT to go 18+

PLUS: Patients control AI and robotics with thought

Zach Mink

October 15, 2025

Good morning, AI enthusiasts. OpenAI is finally breaking the shackles and venturing into the tempting, sometimes even risky, alley of erotica…

CEO Sam Altman says the assistant will soon “treat adult users like adults”. But in the race to make the experience more “enjoyable,” OpenAI may be opening a Pandora’s box — where the line between intimacy and illusion blurs faster than it can be policed.

In today’s AI rundown:

OpenAI to allow erotica on ChatGPT
Ant Group’s thinking model bags IMO silver
Automate market research in minutes
AI slop nears human content on the web
4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

OPENAI

🫣 OpenAI to allow erotica on ChatGPT

Image source: Reve / The Rundown

The Rundown: OpenAI CEO Sam Altman just announced personality improvements planned for ChatGPT, including one that would allow the AI assistant to have mature, erotic conversations with eligible users — much like xAI’s Grok companions.

The details:

Altman said ChatGPT was made restrictive to protect at-risk users, but that hit enjoyment. Now, with age checks and safeguards, OpenAI will ease those limits.
First, it will release a new version of ChatGPT that behaves more like what people liked about 4o, and then by December, it will add support for erotica.
This will allow verified adults to have mature conversations with the AI, giving them “a great deal of freedom” in how they interact with the assistant.
Previously, the company had said it would let devs build “mature” apps that run inside ChatGPT, with appropriate age-gating and controls, The Verge reported.

Why it matters: The AI companion market is huge — and it looks like OpenAI isn’t about to let xAI’s Grok take the lead. While Altman says the mode won’t activate unless users explicitly ask for it, with 800M weekly users, the margin for error is incredibly small, and the risks of emotional dependency on AI companions are not studied.

TOGETHER WITH AIRIA

🛡️ Secure AI, confidently

The Rundown: Airia is the enterprise AI orchestration and security platform that enables organizations to confidently deploy AI with advanced security controls, comprehensive data protection, and automated risk management — ensuring AI initiatives are both powerful and compliant.

With Airia, you'll experience:

Built-in protections for runtime prompts and data leakage
Protection for sensitive data like names, emails, and financial records
Scalable compliance (SOC 2, HIPAA, GDPR)
Granular visibility and control over sensitive data

Discover Airia today.

ANT GROUP

🧠 Ant Group’s thinking model bags IMO silver

Image source: InclusionAI

The Rundown: InclusionAI, Ant Group’s AGI initiative, unveiled Ring-1T, an open-source “thinking model” that internally achieved silver-level on the International Mathematical Olympiad, sitting right behind Google and OAI’s gold-level models.

The details:

Built atop Inclusion’s MoE design, Ring-1T runs on 1T parameters (50B active) with a 128K context window for ultra-long reasoning and multi-turn thinking.
On the IMO test, the model solved four problems in one try and one in three, achieving silver-level and nearly matching Gemini 2.5 Pro on hardest question.
It delivered near SOTA performance across multiple benchmarks, surpassing or closely matching flagship models from Google, Anthropic, and OpenAI.
InclusionAI says Ring-1T is actively being trained and evolving, with goals to unlock deeper reasoning and improve alignment and efficiency.

Why it matters: With near–SOTA performance and almost GPT-5-level reasoning, Ring-1T further blurs the line between open and closed AI. It also marks another major step in China’s bid to challenge the West in the AI race, leaving only a matter of time before the country produces a model capable of taking the global top spot.

AI TRAINING

⚡Automate market research in minutes

The Rundown: In this tutorial, you’ll learn how to use Gemini 2.5 Computer Use to automate market and competitor research, letting AI browse websites, extract pricing and feature data, and compile insights into a structured report.

Step-by-step:

Install Python on your computer (Mac or Windows) to run the Computer Use agent — it’s a one-time setup that takes about five minutes
Clone the Computer Use repository from GitHub, activate the virtual environment, and install dependencies with simple terminal commands (pip install, playwright install)
Get your Gemini API key from AI Studio, link a billing account, and set it as an environment variable to connect the tool to Gemini’s AI model
Run your research task using python main.py --query="YOUR_TASK_INFO", and Gemini will open Chrome, navigate sites, collect pricing and feature data, and output a ready-to-save report.

Pro tip: Run weekly queries to find jobs on job boards like ours (jobs.therundown.ai), track industry trends, research before interviews, and much more.

PRESENTED BY WARP

🏆 Ride with Warp, the world’s No. 1 coding agent

The Rundown: Warp just launched Warp Code – making it the world’s No. 1 coding agent, trusted by over 600,000 developers and 56% of Fortune 500 engineering teams. One platform to prompt, review, and ship production-ready code with agents.

With Warp Code, you get:

New coding capabilities with code review, code editing, in-line code diffs, and file trees
Tops benchmark performance: #1 on Terminal-Bench Verified, 75% on SWE-bench
Access top models: Opus, Sonnet, GPT-5 and more, all in a single subscription
An average of 5 hours saved each week

Download Warp for free to get Pro for only $1 for the first month with code RUNDOWNAI.

AI RESEARCH

🖋️ AI slop nears human content on the web

Image source: Graphite

The Rundown: A Graphite study just found that AI-written articles briefly surpassed human-created ones on the web in late 2024, but the boom has since leveled off, with the web now split roughly evenly between human and AI authors.

The details:

Graphite analyzed 65,000 articles from Common Crawl, published between 2020 and 2025, using Surfer’s AI detector to determine authorship.
The study found that the share of AI-written articles surged after ChatGPT’s launch, peaking well above human output in November 2024.
However, since then, the growth has plateaued, with AI slop staying fairly stable and nearly at the same level as human-written articles.
The researchers attributed this stagnation to the widespread realization that AI-generated content does not perform as well as human content on search.

Why it matters: The great AI content wave appears to be cresting. AI tools can churn out text at scale, but their struggle for visibility is turning much of it into background noise. The findings hint at a new balance (not measured in this study) where human, value-driven content maintains credibility, and AI settles as a collaborator.

QUICK HITS

🛠️ Trending AI Tools

🎉 Encord E-MM1 - The world’s largest multimodal AI dataset is now available open-source with 100m+ images, video, text, audio & point clouds*
🧑‍💻 Perplexity - Now available as a default search engine on Firefox
💻️ Flint - Generate fully-coded on-brand pages on your domain
⚙️ Micro1 - Generate exercises to assess candidates in on-the-job scenarios

*Sponsored listing

📰 Everything else in AI today

Salesforce and OpenAI partnered to integrate Agentforce 360 apps in ChatGPT, enabling direct CRM data and product sales through the assistant’s checkout feature.

Salesforce and Anthropic also expanded their partnership to make Claude the core model for Agentforce 360, develop sector-specific AI solutions, and integrate Claude in Slack.

Walmart is partnering with OpenAI to let ChatGPT users explore and buy its products directly from chat, using the AI assistant’s Instant Checkout feature.

Alibaba's Qwen team introduced more efficient, dense versions of Qwen3-VL 4B and 8B, outperforming models like Gemini 2.5 Flash Lite and GPT-5 Nano.

OpenAI released an updated web search model in Chat Completions, gpt-5-search-api, with 60% lower cost and domain filtering.

Google is launching a Gemini-powered “Help me schedule” feature that suggests meeting time slots based on a user’s Calendar and the context of the email.

COMMUNITY

🤝 Community AI workflows

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader Nancy M. in Cape Charles, Virginia:

"ChatGPT has become my behind-the-scenes creative partner in my art career—it helps me brainstorm new painting ideas, refine product descriptions for my paintings, craft lesson plans for my students, and even polish press releases and outreach emails so my work reaches the right audience.”

How do you use AI? Tell us here.