OpenAI supercharges ChatGPT: New Canvas and Projects features unveiled: All details
Announced via OpenAI's official X account, the update brings long-requested features to Canvas — a live editing and formatting space within ChatGPT, initially launched in preview form in October 2024 and later rolled out to all users by December. With this update, users can now export their work in several formats, including PDF, Microsoft Word (.docx), and Markdown (.md). For developers using Canvas to write code, direct export in language-specific formats such as Python (.py), JavaScript (.js), and SQL (.sql) is now possible.
The Canvas feature is accessible across all platforms where ChatGPT is available — including the website, desktop clients, and mobile apps — allowing users to maintain a consistent workflow regardless of device.
In parallel, OpenAI has announced substantial upgrades to the Projects tool, which is exclusive to ChatGPT's paid subscribers. Projects allows users to organise conversations and store contextual information, including previous chats, uploaded files, and personalised instructions. The new update significantly extends its scope.
Most notably, Projects now supports 'deep research' capabilities. This allows users to make complex, multi-step research queries using real-time data from public web sources, improving the AI's performance in tasks that demand more than surface-level information. Additionally, the integration of ChatGPT's voice mode provides a hands-free experience, making it easier for users to interact with the assistant on the go.
Further improvements are also being introduced on mobile. Users of the ChatGPT app on iOS and Android can now upload files and switch between AI models directly within a project — two features previously limited to the desktop experience.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Indian Express
13 minutes ago
- Indian Express
‘I stopped being afraid of my numbers': US woman pays off $12K credit card debt with ChatGPT's help
A 35-year-old woman from Delaware is inspiring thousands online after revealing how she paid off more than $12,000 (Rs. 10.28 lakh, approximately) of her $23,000 (Rs. 19.7 lakh, approximately) credit card debt in just 30 days with the help of ChatGPT. Jennifer Allen, a realtor and content creator, told Newsweek that although she earned a decent income, managing her finances had always been a struggle. 'Not because I don't make enough, but because I was never taught financial literacy,' she admitted. For years, she focused on working harder instead of tackling the real issue that was 'poor financial habits'. Things came to a head after she became a mother. 'We weren't living lavishly, we were just surviving. But the debt piled up while I wasn't looking,' she said. To take control of her finances, Jennifer gave herself a 30-day challenge and used ChatGPT as her guide. Each day, she asked the AI tool for simple, practical financial advice and made a point to follow through. The suggestions were simple and straightforward, and the key was in the consistency. 'It wasn't some big financial hack,' Jennifer said, adding, 'It was the act of facing it every day, tracking it, talking about it, looking at it. I stopped being afraid of my numbers.' The AI suggested steps like cancelling unused subscriptions, selling household items on Facebook Marketplace, and planning meals using what was already in her kitchen. One of the biggest wins came when ChatGPT advised her to review her dormant accounts, leading her to discover a brokerage account she had forgotten about, with $10,000 (Rs. 8.56 lakh, approximately) in it. Another money-saving tip came when ChatGPT recommended cutting back on grocery expenses by using existing pantry items. Jennifer said that the strategy alone saved her hundreds of dollars. At the end of the 30 days, she had paid off more than half of her debt. Encouraged by her progress, Jennifer now plans to do a second 30-day challenge to clear the remaining balance. Allen gave simple advice to people by simply stating, 'Don't wait until you feel ready or smart enough to start. You don't have to know all the answers. You just have to stop pretending it's not happening.'
Mint
an hour ago
- Mint
7 simple prompts that can make AI chatbots more useful in daily life
Today, it is very clear that artificial intelligence is part of both workplaces and households. Tools like ChatGPT, Claude, and Gemini are used by many people, but how helpful they are often depends on how clear the prompts are. While most people use them for short replies or quick searches, giving more specific instructions can help these chatbots handle daily tasks more reliably. A recent survey by the Digital Work Lab found that almost 60% of professionals use AI chatbots every week. Many said they started with simple questions but soon noticed that clear prompts gave more helpful answers. The examples below show seven easy prompts that work well in daily situations. Here are six little known facts that can help you get more out of your chatbot, along with seven smart prompts you may want to try right away. 1) They respond better to precise instructions about style and length Many people start with short or general questions and receive broad answers. When you include details such as the tone, word count, or format, chatbots often produce clearer replies. For example, asking to summarise a text in no more than 100 words usually results in a more focused response. This way, you are more likely to get accurate and clear replies. 2) They don't think in words the way people do Chatbots break text into small pieces called tokens rather than processing whole words. This means even small changes in your prompt, like adding punctuation or rephrasing a sentence, can lead to very different answers. For example, 'Explain AI simply' may produce a shorter reply than 'Can you please explain AI simply?' because the tokens are arranged differently. 3) They can create several versions in one reply Instead of repeating the same request, you can ask for different versions in a single prompt. A question like writing three variations of an email, one formal, one casual, and one short, gives options to choose from without extra steps. 4) Chatbots don't always have the most up-to-date information Most chatbots were trained on data that stops at a certain date. If you ask about something new, such as a recent product release or a viral trend, the response could be outdated unless web browsing is turned on. Example: If you type, 'Who won the 2025 election?' the chatbot may respond that it doesn't have that information or that its data ends in 2024. You can also add instructions such as 'Check for recent information' or 'Look online if available' to help the chatbot look for the latest details. 5) Negative instructions can help simplify responses When you say what you don't want in the answer, the chatbot often gives a simpler reply. For example, asking it to leave out technical words can make the explanation easier to understand. 6) Tone typically reflects the user's prompt Chatbots tend to adopt the style set by the person writing the question. They will match a friendly, casual tone or maintain a formal voice if requested at the start. By now, you must have got a fair idea of how to use chatbots. These seven prompts can help you save time, get clearer answers, and handle daily tasks more easily with tools like ChatGPT, Claude, and Gemini. 1) Create a learning plan You can ask the chatbot to set up a 14 day plan for learning a new skill step by step. This helps you stay organised and know exactly what to work on each day. Create a 14 day plan to learn basic Excel, including daily tasks and helpful resources. 2) Write a polite email declining a meeting The chatbot can draft a clear, polite message saying you can't join a meeting and suggesting a new time. This saves you time and keeps your communication professional. Write a polite email to decline a meeting on Friday and suggest next Wednesday instead. 3) Summarise long text in bullet points You can paste a long text and get a short version in bullet points. This makes it much easier to read and share with others. Summarise this document in 8 bullet points under 150 words. 4) Identify current search trends The chatbot can list popular keywords in any topic so you know what people are searching for right now. This helps you pick topics that are more likely to get attention. List trending keywords for health and wellness this month. 5) Compare two products side by side You can ask for a table showing the features, prices, and pros and cons of two products. This helps you see differences clearly before you decide what to buy. Compare iPhone 16 and OnePlus 13 in a table with features, prices, pros, and cons. 6) Suggest event ideas on a limited budget The chatbot can suggest ideas for a team event that fits within a set budget. This makes planning easier without spending too much. Suggest 5 ideas for a team event in India with a budget under ₹ 20,000. 7) Explain a complex topic in simple terms You can ask the chatbot to explain a topic in plain language without hard words. This helps you understand something new without feeling confused. Explain blockchain in plain language without using technical words. Chatbots do a lot of good, but they aren't perfect. The more you learn about how they work, the easier it is to use them without confusion and get better results.
Time of India
an hour ago
- Time of India
Do not trust ChatGPT, it can hallucinate, says OpenAI CEO Sam Altman: Why students must rethink AI dependence
Artificial intelligence, an innovation that began attracting a plethora of attention while still in its cradle, is now growing up in a world determined to dominate workplaces. It has made inroads into every aspect of human existence. And education is no exception. In fact, one could argue that the most treacherous illusion in modern education is not misinformation, but misplaced trust. This technology has not only revolutionised the workplace—it has also weakened the very foundation of our education system. A 2023 Wall Street Journal survey revealed that approximately 90% of students were using ChatGPT to complete their assignments. The technology has introduced an array of shortcuts and instant answers, serving them up on a silver platter. But what happens when the very tool designed to enhance thinking quietly replaces it? That question is no longer hypothetical; it's an unsettling reality of our present context. ChatGPT is not just disrupting the education system; it is overwhelming students with misinformation. But who's saying this? A critic? Not quite. It's the creator himself. Yes, you read that right. In a rare moment of candor, OpenAI CEO Sam Altman , the architect behind ChatGPT, issued a caution: 'People have a very high degree of trust in ChatGPT, which is interesting, because AI hallucinates. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Τι είναι το ChatGPT για το οποίο μιλάνε όλοι; courses AI Undo It should be the tech you don't trust that much.' The warning demands immediate introspection, especially from students who are increasingly treating generative AI as gospel. This isn't paranoia. It's a reckoning. The authority of the algorithm, and its limitations Altman's warning came during the inaugural episode of OpenAI's official podcast, where he reflected on ChatGPT's growing influence. While acknowledging its evolving capabilities, he admitted the technology is still 'not super reliable'. The irony couldn't be louder: A CEO asking users not to trust his own flagship product. Why? Because ChatGPT doesn't know. It predicts. The chatbot crafts responses based on probability, not truth. It mimics understanding without possessing it. And in academic settings, where nuance, originality, and interpretation matter more than fluency, that can be a silent killer of intellectual development. Altman's words arrive amid mounting legal pressure, from copyright lawsuits to privacy concerns, and a broader realization that generative AI, though dazzling, is still deeply flawed. The platforms remoulding how students learn, research, and write are also reprogramming how they think, or whether they are thinking at all. A generation at risk of intellectual erosion The danger here isn't just factual inaccuracy. It's cognitive atrophy. Students, allured by ChatGPT's eloquence and speed, often bypass the process of struggling through a problem. That struggle, frustrating as it may be, is where learning lives. Without it, the brain stops forging new pathways. It memorizes outputs, but forgets how to arrive at them. True education is not glittering degrees, but developing the ability to think, to form opinions, and to have knowledge of certain events. Critical thinking, curiosity, and deep reflection lie at the heart of true education, are now standing on the edge of becoming obsolete in a world that rewards quick completions over complex contemplations. Reversing the decline: What students must do now Altman's admission isn't a condemnation of AI. It's a provocation. One that calls for a cultural reset in how students engage with this tool. Here's how students can begin reclaiming their intellectual agency: Distrust, then verify : Use AI as a spark, not scripture. Always fact-check, especially on historical, legal, or scientific topics. Think before you prompt : Engage your own ideas first. Let ChatGPT refine—not replace—your thinking. Own your authorship : AI can help draft. Only you can craft. Your voice matters more than its fluency. Understand its boundaries : Generative AI does not reason. It does not know you. Don't assign it a wisdom it doesn't possess. Use it with conscience : Relying on AI to do your thinking isn't a shortcut; it's a slow exit from your own mental independence. Final thought: A mirror, not a mentor In the rush to innovate, we risk creating machines that write essays no one remembers, solve equations no one understands, and provide answers no one questions. The greatest threat of AI isn't misinformation, it's intellectual complacency. Sam Altman's unexpected honesty is not just a cautionary footnote in tech history. It's a line in the sand for educators, students, and institutions alike. Let ChatGPT be a tool, not a tutor. Let it assist, not define. And above all, let it remind us that the most powerful engine of learning still lies between our ears, not inside a prompt box. Is your child ready for the careers of tomorrow? Enroll now and take advantage of our early bird offer! Spaces are limited.
