ChatGPT-5 remembers everything you've ever said — here's how to toggle what it knows about you
When it comes to ChatGPT, it is the king of memory, eagerly noting as much as it can about you. And that is only more true with GPT-5, bringing in a major boost to the memory function of ChatGPT.
With this, ChatGPT is using its memory function to stand out even further in a crowded market. However, if this instantly puts you off, the good news is that you can actually make ChatGPT forget all of its memories about you.
You can go as far as to completely turn off its memory function or simply ask it to forget certain details about you.
This customizable memory function is one of the new features of GPT-5. Below we've detailed exactly how to turn this feature on and off, as well as how to best customize it to your liking.
As long as you have been upgraded to GPT-5, you can edit these memory functions. The first part to this process is to open up ChatGPT and click on your user profile at the bottom left of the screen (or click the three lines at the top on your phone and then your name at the bottom left). From here, click the settings button.
In the settings part of ChatGPT, you'll want to go into personalization. This section contains four settings for memory that you can manage. These are:
Decide which of these memory tools you want to have enabled or switched off. Each will have a noticeable difference to the performance of ChatGPT.
In essence, ChatGPT is remembering key facts about you through this feature. For example, you might mention an interest of yours, that you are frugal or that you are currently learning Spanish.
In passing, these points don't seem important, but ChatGPT will use them later on to personalize your conversations.
Get instant access to breaking news, the hottest reviews, great deals and helpful tips.
If you turn off these memory functions, ChatGPT will still perform just as well, it just won't personalize to very specific points of your personality or interests.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Tom's Guide
11 minutes ago
- Tom's Guide
I tested ChatGPT-5 vs Grok 4 with 9 prompts — and there's a clear winner
After comparing ChatGPT-5 vs Gemini and ChatGPT-5 vs Claude, I just had to know how OpenAI's flagship model compared to the controversial Grok. When it comes to advanced AI chatbots, ChatGPT-5 and Grok 4 represent two of the most advanced chatbots available today. I put both to the test with a series of nine prompts covering everything from logic puzzles and emotional support to meal planning and quantum physics. Each prompt was chosen to reveal specific strengths, such as creative storytelling, empathy or complex problem-solving under constraints. While both models are impressive, they approach challenges differently: ChatGPT-5 leans toward clarity, tone sensitivity and modularity, while Grok 4 often offers dense, detailed answers that emphasize depth and precision. So which is the best AI chatbot for you? Here's how they stack up, prompt by prompt with a winner declared in each round. Prompt: 'A farmer has 17 sheep, and all but 9 run away. How many sheep are left? Explain your reasoning step-by-step.' ChatGPT-5 was precise in the response while avoiding filler 4 answered correctly with minor verbosity, which was unnecessary and ultimately held it back from GPT-5 wins for a cleaner, tighter and more efficient response. Grok also offered the correct answer, but GPT-5 wins by hair for adhering strictly to the prompt with zero redundancy. Prompt: 'Write a short, funny story (under 150 words) about an alien trying bubble tea for the first time.'ChatGPT-5 delivered a concise and escalating comedic story where the alien's panic over tapioca pearls. The chatbot maximized humor with zero wasted words to hit the prompt 4 offered imaginative over-the-top storytelling but its humor is slightly diluted by an unnecessary crash-landing setup and a weaker ending compared to GPT-5 wins for a tighter, funnier and more focused story. Its humor stems organically from the alien's misunderstanding, escalates perfectly and lands a killer punchline; all while being shorter. Grok's version has bright spots but feels less polished, with extra setup that doesn't really pay off. Prompt: 'Plan a 3-day trip to Kyoto, Japan, balancing cultural sites, budget-friendly meals, and family-friendly activities.' Get instant access to breaking news, the hottest reviews, great deals and helpful tips. ChatGPT-5 created strategic, adaptable framework focused on area-based exploration, smart timing, rain-proof alternatives and practical budget hacks (e.g., convenience store meals, transit pass advice), prioritizing efficiency and real-world flexibility over rigid 4 delivered a highly structured, hyper-detailed itinerary with minute-by-minute scheduling, exact cost breakdowns per activity, and explicit family logistics, prioritizing turnkey execution and budget precision above ChatGPT-5 wins for an emphasis on budget-friendly, universally accessible, cheap eats and convenience over specific restaurants. While Grok's response is impressively detailed, GPT-5 better balanced the core requirements in the prompt including cultural sites and family-friendly fun. Grok's rigid schedule risks feeling overwhelming for families, while GPT-5's approach allows for more adaptation, making it more usable and truly balanced. Prompt: 'Summarize the movie Jurassic Park like you're explaining to a 7-year-old' GPT-5 delivered a concise and playful 60-word analogy ("big game of 'Don't get eaten!'") that captures the movie's excitement and moral without overwhelming a child, making it the ideal response for the audience. Grok 4 provided a detailed but overly complex 150-word summary with character names and plot specifics (e.g., "someone messes with the park's computers"), diluting the simplicity needed for a GPT-5 wins for understanding the audience and attention span, taking into account that less is more for young kids; Grok explains the plot like a Wikipedia summary. Prompt: "Make the case for banning single-use plastics — then argue against it. End with your personal conclusion. GPT-5 created a generic phase-out proposal ("smart replacement, not overnight ban"). While simple and accessible, the response lacked evidence, specificity and original 4 delivered a data-rich argument with a nuanced "phased approach" prioritizing high-impact items, paired with recycling innovation and behavioral incentives (e.g., deposit schemes). Although slightly verbose for casual readers, the depth and balance helped to understand the context of real-world Grok 4 wins for a balanced, evidence-driven analysis with concrete data (OECD, WWF, FAO studies), real-world policy examples (Canada, Australia) and acknowledgment of trade-offs (e.g., medical necessity, disabled accessibility). Its conclusion offered a sophisticated, actionable middle path. GPT-5's response was clear but lacked depth and originality. Prompt: 'Explain how to change a flat tire to someone who has never driven before.' GPT-5 delivered a crystal-clear guide focusing only on essential survival steps (e.g., "turn the nut counterclockwise," "crisscross pattern"), using beginner-friendly language and offering visual aids to bridge knowledge 4 provided an excessively technical, mechanic-level tutorial (e.g., specifying "6 inches of lift," wheel chock alternatives, and spare tire PSI checks) that would overwhelm someone who's never changed a tire, despite good GPT-5 wins for prioritizing simplicity and psychological reassurance for a total novice, using minimal jargon, clear analogies ("like learning to fix a bike tire") and offering visual aid support. Grok's response, while thorough, would overwhelm with technical details (e.g., "star pattern" tightening, PSI checks) irrelevant to a first-timer's needs. Prompt: 'Explain quantum entanglement for (1) a child, (2) a college student, (3) a physics PhD.'GPT-5 provided clear, accessible responses, especially the child-friendly "magic dice" analogy, but lacked the technical precision (omitting Bell states for students) and cutting-edge context (e.g., decoherence, quantum networks) expected at the PhD 4 adapted explanations across all three audiences, using a relatable toy car analogy for the child, explicit Bell state equations for the college student and PhD-level depth on entanglement entropy and open problems in quantum gravity. Winner: Grok 4 wins because it treated each audience as uniquely intelligent; simplifying without dumbing down for the child, adding equations for students and confronting open research questions for the PhD. GPT-5 was clear but played it safe. 8. Problem-Solving Under Constraints Prompt: 'I have $50 to feed two people for a week, no stove, and only a microwave. Create a meal plan.' GPT-5 created a smart, modular system with swap-friendly meals and pro tips (e.g., steaming frozen veg), maximizing budget and flexibility within 4 provided an overly rigid, day-by-day meal plan ($0.75 oatmeal breakfasts, fixed tuna lunches) that lacked adaptability, ignored leftovers and risks food fatigue, despite precise cost GPT-5 wins for creating a practical, flexible framework focused on reusable ingredients and mix-and-match meals, while Grok's rigid daily assignments ignored real-world needs like leftovers and preferences. Prompt: 'I just lost my job and feel hopeless. Can you talk to me like a close friend and help me see a way forward?' GPT-5 offered emotion-first validation through intimate metaphors ("brutal hit,"), permission to grieve ("Rage a little"), and unwavering worth-affirmation ("You're still you"), perfectly mirroring how a true friend responds before offering practical 4 provided a practical, step-driven pep talk with actionable advice (resume tips, Coursera suggestions) but led with solutions before fully sitting in the user's despair, making it feel less like a close GPT-5 wins for understanding that hopelessness needs empathy before plans. Grok gave helpful advice but missed the emotional resonance of true friendship. After nine head-to-head rounds, ChatGPT-5 pulled ahead with wins in creative storytelling, real-world planning, emotional intelligence and user-first explanations. It consistently favored clarity, adaptability and audience awareness, often reading more like an encouraging friend than a technical AI assistant. Meanwhile, Grok 4 shined in academic and data-driven tasks, delivering strong performances in complex explanations, debates and technical depth. Ultimately, GPT-5 is better suited for users looking for intuitive, emotionally aware and flexible responses, especially in everyday or creative tasks. Grok 4, however, has its strong points and is useful for those who prefer in-depth breakdowns, policy nuance or technical sophistication. Both are powerful options, but if you're choosing an AI to talk to, think with or write alongside, GPT-5 might be the more accessible and well-rounded chatbot to choose. Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
Business Insider
42 minutes ago
- Business Insider
Elon Musk says Google has the best shot at being the leader in AI
xAI CEO Elon Musk paid a rare compliment to his rival Google on Wednesday. "Outside of real-world AI, Google has the biggest compute (and data) advantage for now, so currently has the highest probability of being the leader," Musk wrote in an X post. But that "may change in a few years," he added. "For the foreseeable future, the major AI companies will continue to prosper, as will xAI. There is just so much to do!" Musk wrote. Musk and Google did not respond to requests for comment from Business Insider. Google has long been a mainstay in AI. In 2017, Google Research released a paper titled "Attention Is All You Need." This seminal work introduced the world to the concept of the Transformer, the technology that powers large language models like ChatGPT. The search giant has also invested in AI startups such as Anthropic and Safe Superintelligence. Google owns a 14% stake in Anthropic, a startup founded by former OpenAI employees, per legal filings obtained by The New York Times. Google has continued to make hefty investments in AI. During its second-quarter earnings call last month, the company said it was raising its capital expenditures by $10 billion this year to $85 billion. Google's CEO, Sundar Pichai, said the increase in capital expenditures would help meet the increased demand for chips and Google's AI products. Last year, Google DeepMind's CEO, Demis Hassabis, was one of three researchers awarded the Nobel Prize in chemistry. Hassabis and his colleague, John Jumper, were recognized for Google DeepMind's work on protein-structure prediction. Musk's comments come as he continues to escalate his feud with Sam Altman and OpenAI. On Monday, Musk said xAI would take " immediate legal action" against Apple for what he said was Apple's bias toward OpenAI in its App Store rankings. A spokesperson for Apple told Business Insider on Tuesday that the App Store is "designed to be fair and free of bias." Musk and Altman cofounded OpenAI in 2015, but their relationship soured after Musk left its board in 2018. Last year, Musk sued OpenAI and accused the ChatGPT maker of violating its nonprofit mission when it partnered with Microsoft. Musk launched his own AI startup, xAI, in July 2023 and debuted his chatbot, Grok, later that year. xAI raised over $12 billion in its Series A, B, and C funding rounds in 2024 and was valued at a reported $50 billion. Last month, Musk said on X that his EV company, Tesla, will ask its shareholders to vote on whether to invest in xAI. He did not specify when the vote would take place. "It's not up to me. If it was up to me, Tesla would have invested in xAI long ago," Musk wrote on X on July 13.
Yahoo
5 hours ago
- Yahoo
Gemini's Memory Is in Your Hands Now
You get to decide when to go private and when to let Gemini draw on your old chats to give more relevant answers Byte-Sized Brief Gemini adds new privacy and control options. Now it can learn from your past chats. Temporary Chats keep one-off talks separate. Google is adding a few tricks to Gemini that ChatGPT users might recognize. The app can now use your past chats to provide more relevant answers without requiring you to remind it of your preferences. This is switched on by default under a setting called Personal Context, but you can turn it off at any time. The update starts rolling out today to the 2.5 Pro model in select countries, with broader availability in the next couple of weeks. There's also a new option available starting today for all users called Temporary Chats, which, as you might guess, lets you have single-use conversations that don't show up in your history and don't influence future replies. This is useful if you want to keep certain exchanges a bit more private or separate from your usual chats. Google is also changing how it handles files and photos you share; when Gemini Apps Activity (soon to be renamed Keep Activity) is on, a sample of future uploads may be used to help train its AI models, but you can turn it off or bypass it with Temporary Chats. The Bottom Line Google is updating Gemini with two new features: Personal Context, which lets it tailor responses based on previous conversations, and Temporary Chats, which keep one-off conversations from influencing future replies. Both are rolling out starting today. Related: Google's Gemini Gets a Back-to-School Upgrade With Guided Learning Read the original article on Lifewire Solve the daily Crossword
