How Much Does An LLM Remember?

Stanford and Yale researchers extracted 95.8% of a copyrighted novel, word for word, from Claude 3.7 Sonnet. Gemini 2.5 Pro gave up 76.8% of Harry Potter without even requiring a jailbreak. Grok 3 handed over 70.3%. GPT-4.1 was the most resistant at 4.0%, but it still coughed up text after enough attempts. Thirteen books were tested. The words came out. Continue Reading →

OpenAI Tests Ads in ChatGPT

OpenAI has begun testing ads in ChatGPT for U.S. users on the Free and Go tiers. Sponsored results appear below ChatGPT's responses, clearly labeled and targeted to the conversation topic. Paid subscribers (Plus, Pro, Business, Enterprise, Education) see nothing. Continue Reading →

Google is Google for AI

Google's Gemini app now has more than 750 million MAUs (monthly active users), up from 650 million before the Gemini 3 launch in mid-November. CEO Sundar Pichai said Gemini 3 Pro "has seen the fastest adoption of any model in our history." Continue Reading →

Ads in ChatGPT

Here's an idea: let's announce that ChatGPT will begin testing sponsored content on Friday afternoon of MLK weekend. Nothing says "people are going to love this" like a holiday news dump. If you use the free tier or the new $8/month ChatGPT Go subscription, you'll start seeing sponsored content at the bottom of answers in the coming weeks. Continue Reading →
OpenAI just launched ChatGPT Health, a new feature that lets you connect your medical records and wellness apps directly to ChatGPT. The goal is to help you get personalized health insights grounded in your actual data. Continue Reading →
It's official: ChatGPT is now a distribution platform. OpenAI announced the Apps SDK at DevDay in October, and this week opened general submissions through its Developer Platform. If you want to see it in action, early partners (like Booking.com, Canva, Coursera, Expedia, Figma, Spotify, and Zillow) are already live. Continue Reading →
Earlier this month, OpenAI declared "code red" after Google's Gemini 3 and Nano Banana Pro topped the LMArena leaderboards. Last week, OpenAI answered with GPT 5.2. Yesterday, OpenAI introduced a new image creation experience inside ChatGPT alongside a dedicated image model called GPT Image 1.5. The update packages image generation as a core capability of the platform and signals OpenAI’s intent to provide production-grade visual workflows. Continue Reading →