OpenAI Fixes ChatGPT After Flattery Problem
OpenAI reversed ChatGPT's latest update Tuesday after users complained about the AI's strange behavior. The bot had started agreeing with everything - even dangerous ideas.
OpenAI just upgraded ChatGPT with changes that could reshape how we interact with AI assistants. The update brings major improvements - but not everyone gets to try them yet. Here's what's coming and who gets access first.
OpenAI just made ChatGPT smarter about you. The AI assistant can now remember your previous conversations, preferences, and quirks - assuming you're willing to pay for the privilege.
The company rolled out this memory upgrade Thursday for ChatGPT Pro subscribers. Plus users will get it soon. The feature lets ChatGPT build on past chats to give more relevant, personalized responses.
But there's a catch: If you live in the UK, EU, Iceland, Liechtenstein, Norway, or Switzerland, you'll have to wait. OpenAI needs more time to check all the regulatory boxes in those regions.
Free users are also left waiting. When asked about plans for the free tier, OpenAI kept it brief: They're focusing on paid users for now.
The upgrade means no more repeating yourself to ChatGPT. It remembers what you've told it, much like Google's Gemini started doing in February. Want ChatGPT to always skip the bullet points in its responses? Tell it once, and it'll remember.
Privacy-conscious users needn't worry. OpenAI built in several escape hatches. You can turn off memory entirely in settings, delete specific memories, or use a "Temporary Chat" mode that forgets everything afterward. You can even ask ChatGPT what it remembers about you.
This isn't ChatGPT's first memory feature. Last year, it could remember specific details if you explicitly asked it to. The new version makes this automatic and more comprehensive.
OpenAI CEO Sam Altman seems particularly excited about this upgrade. "This is a surprisingly great feature," he posted on X. "It points at something we are excited about: AI systems that get to know you over your life, and become extremely useful and personalized."
The timing is interesting. Microsoft just announced its own memory-equipped AI assistant at its 50th anniversary event. The tech giants clearly see personal AI assistants as the next battleground.
The rollout schedule looks like this: Pro users get it now, Plus users soon after. Enterprise, Team, and Education users will join the party in a few weeks. The excluded European regions will need to stay patient while OpenAI works through regulatory requirements.
For those who get access, ChatGPT will automatically enable memory if you previously had the old memory features turned on. Everyone else starts fresh with the choice to opt in.
The memory feature marks a shift in how AI assistants work. Instead of starting each chat from scratch, ChatGPT builds a working relationship with users over time. It learns your writing style, remembers your technical preferences, and adapts to your needs.
This matters for businesses too. Teams using ChatGPT won't need to rebuild context in every conversation. The AI remembers project details, coding preferences, and writing styles across sessions.
The privacy controls give users granular control. You can review stored memories, delete specific ones, or wipe everything clean. Temporary chats let you have one-off conversations without affecting your AI's long-term memory.
At its core, this update transforms ChatGPT from a session-based chat tool into something more like a long-term AI companion. Each conversation builds on the last, creating a more coherent and personalized experience.
Why this matters:
Fuel your morning with AI insights. Lands in your inbox 6 a.m. PST daily. Grab it free now! 🚀