OpenAI announced group chats as a ChatGPT feature, allowing users in select regions to collaborate with each other — and with the AI — in a shared conversation for planning, decision-making or working through ideas. In fact, many users have reported instances of GPT-5 making simple mistakes. Despite these clear advances compared to prior models, GPT-5 is still susceptible to the same problems that plagued its predecessors. The release of GPT-5 provides a sizable update to previous models, at least on paper.
How Does ChatGPT Work?
This year alone, OpenAI peppered us with new models, modes, and features to play with inside the AI. Some, including Nature and JAMA Network, “require that authors disclose the use of text-generating tools and ban listing a large language model (LLM) such as ChatGPT as a co-author”. As of 2023, there were several pending U.S. lawsuits challenging the use of copyrighted data to train AI models, with defendants arguing that this falls under fair use. A shadow market has emerged for users to get access to foreign software tools. ChatGPT has never been publicly available in China because OpenAI prevented Chinese users from accessing their site.
Deep Research
- Team and Enterprise plans are designed not to collect user data by default.
- “You can create systems that can fool people into thinking that this thing that you’re talking to is human,” Jaffri tells CNET.
- AI agents represent stage three of AGI, enhancing ChatGPT’s reasoning abilities and enabling it to complete a broader range of tasks without human assistance.
- To build a safety system against harmful content (e.g., sexual abuse, violence, racism, sexism), OpenAI used outsourced Kenyan workers earning around $1.32 to $2 per hour to label such content.
It found that 52% of the responses contained inaccuracies and 77% were verbose. One study analyzed ChatGPT’s responses to 517 questions about software engineering or computer programming posed on Stack Overflow for correctness, consistency, comprehensiveness, and concision. Robin Bauwens, an assistant professor at Tilburg University, found that a ChatGPT-generated peer review report on his article mentioned nonexistent studies. In January 2023, Science “completely banned” LLM-generated text in all its journals; however, this policy was just to give the community time to decide what acceptable use looks like.
Select a mode for your prompt
OpenAI acknowledged that there have been “instances where our 4o model fell short in recognizing signs of delusion or emotional dependency”, and reported that it is working to improve safety. In July 2025, OpenAI released ChatGPT agent, an AI agent that can perform multi-step tasks. It is capable of writing software, answering codebase questions, running tests, and proposing pull requests. According to TechCrunch, it is a service based on o3 that combines advanced reasoning and web search capabilities to make comprehensive reports within 5 to 30 minutes. This tool allows a user to customize ChatGPT’s behavior for a specific use case. ChatGPT’s Mandarin Chinese abilities were lauded, but the ability of the AI to produce content in Mandarin Chinese in a Taiwanese accent was found to be “less than ideal” due to differences between mainland Mandarin Chinese and Taiwanese Mandarin.
Regional responses
In July 2023, the FTC launched an investigation into OpenAI, the creator of ChatGPT, over allegations that the company scraped public data and published false and defamatory information. In late March 2023, the Italian data protection authority banned ChatGPT in Italy and opened an investigation. Stanford researchers reported that GPT-4 “passes a rigorous Turing test, diverging from average human behavior chiefly to be more cooperative.” A 2023 study reported that GPT-4 obtained a better score than 99% of humans on the Torrance Tests of Creative Thinking. The company announced a slew of generative AI-powered features to counter OpenAI and Microsoft.
Between March and April 2023, Il Foglio published one ChatGPT-generated article a day on its website, hosting a special contest for its readers in the process. The chatbot technology can improve security by cyber defense automation, threat intelligence, attack identification, and reporting. In an industry survey, cybersecurity professionals argued that it was attributable to cybercriminals’ increased use of generative artificial intelligence (including ChatGPT).
Story problem
Based on internal evaluations, GPT-5 outperforms its predecessors in multiple areas, including math, coding, visual perception and health knowledge. OpenAI released GPT-5, calling it its “fastest” and “smartest” AI model to date. The company also unveiled an open-source technical standard called Agentic Commerce Protocol for merchants to build integrations with ChatGPT, essentially making their products available for purchase within the chatbot.
- Let’s not forget when OpenAI’s “Sky” voice model mimicked Scarlett Johansson without permission, prompting her team to threaten legal action over the unauthorized use of her voice.
- In late March 2023, the Italian data protection authority banned ChatGPT in Italy and opened an investigation.
- This often leads to what experts call “hallucinations,” where the output generated is stylistically correct, but factually wrong.
- These include, among many others, writing and debugging computer programs, composing music, scripts, fairy tales, and essays, answering questions (sometimes at a level exceeding that of an average human test-taker), and generating business concepts.
- Advanced voice mode becomes available on web browsers — before, it could only be accessed on mobile and desktop formats.
In March 2025, OpenAI updated ChatGPT to generate images using GPT-4o instead of DALL-E. ChatGPT is frequently used for translation and summarization tasks, and can simulate interactive environments such as a Linux terminal, a multi-user chat room, or simple text-based games such as tic-tac-toe. It is designed to generate human-like text and can carry out a wide variety of tasks.
It was capable of autonomously performing tasks through web browser interactions, including filling forms, placing online orders, scheduling appointments, vegas casino download and other browser-based tasks. At launch, OpenAI included more than 3 million GPTs created by GPT Builder users in the GPT Store. In March 2023, ChatGPT Plus users got access to third-party plugins and a browsing mode (with Internet access). Premium users were originally limited in the number of messages they could send to the new model, but OpenAI increased and eventually removed these limits.
At launch, the feature was limited to purchases on Etsy from US users with a payment method linked to their OpenAI account. The user can interrupt tasks or provide additional instructions as needed. In 2025, OpenAI added several features to make ChatGPT more agentic (capable of autonomously performing longer tasks). The integration used ChatGPT to write prompts for DALL-E guided by conversations with users.
There are different model variations and limitations based on your subscription and your purpose for using ChatGPT. If you prefer to keep using Dall-E 3 (its first image generator), you can still access it through the Dall-E GPT. ChatGPT Canvas can help you write and code, and with the Tasks feature, you can set up reminders and schedule tasks. It won’t appear in history, update its memory or be used to train models, although it may be kept for up to 30 days for safety. It also rolled GPT-4o back to its prior version after users on X complained that its “sycophantic” praise felt unsettling and distressing.
This iteration is multimodal, meaning it can process text, images and audio. In March 2023, OpenAI announced GPT-4, the latest version of its language model. Through its ChatGPT Search feature, it is competing more directly with Google search (even removing mandatory login requirements to make it that much easier for casual users). OpenAI expanded ChatGPT’s memory feature, allowing the chatbot to recall previous interactions (which you can manage or delete), creating a more personalized user experience. OpenAI did not respond to a request for comment on potential copyright and plagiarism concerns about training data or on user privacy and data protection.
Microsoft’s new AI agents create your Word, Excel, and PowerPoint projects now
Custom instructions allow users to save directions that apply to all interactions, rather than adding them to every request. According to an OpenAI blog post, ChatGPT will build memories on its own over time, though users can also prompt the bot to remember specific details — or forget them. To kick off the start of its o1 model family, OpenAI also announces the limited release of its o1-preview and o1-mini models. OpenAI initially released this feature to users who pay for the Plus, Team or Enterprise plan. The feature is also still exclusive to ChatGPT users who have a Plus, Team, Enterprise or Education plan.
Another study, focused on the performance of GPT-3.5 and GPT-4 between March and June 2024, found that performance on objective tasks like identifying prime numbers and generating executable code was highly variable. Chris Granatino, a librarian at Seattle University, noted that while ChatGPT can generate content that seemingly includes legitimate citations, in most cases those citations are not real or largely incorrect. As of July 2025, Science expects authors to release in full how AI-generated content is used and made in their work. Over 20,000 signatories including Yoshua Bengio, Elon Musk, and Apple co-founder Steve Wozniak, signed a March 2023 open letter calling for an immediate pause of giant AI experiments like ChatGPT, citing “profound risks to society and humanity”. OpenAI said it has taken steps to effectively clarify and address the issues raised; an age verification tool was implemented to ensure users are at least 13 years old.
The plan costs $200 per month and gives users access to Open AI o1, o1-mini, GPT-4o and advanced voice mode. The new feature enables the chatbot to start and complete tasks on behalf of the user with minimal input. Through the partnership, ChatGPT users will be able to find products sold by Walmart and complete entire purchases without leaving the chatbot’s interface. The partnership also involves Intuit deepening its use of OpenAI’s frontier models to support a range of use cases on its platform, including powering select AI agents and improving employee productivity at Intuit. “We’ll use language models to help us write first drafts, to brainstorm.
ChatGPT shows inconsistent responses, lack of specificity, lack of control over patient data, and a limited ability to take additional context (such as regional variations) into consideration. The ChatGPT-generated avatar told the people, “Dear friends, it is an honor for me to stand here and preach to you as the first artificial intelligence at this year’s convention of Protestants in Germany”. In July 2024, Futurism reported that GPT-4o in ChatGPT would sometimes link “scam news sites that deluge the user with fake software updates and virus warnings”; these pop-ups can be used to coerce users into downloading malware or potentially unwanted programs.
Released in February 2025, GPT-4.5 was described by Altman as a “giant, expensive model”. In September 2024, OpenAI introduced o1-preview and a faster, cheaper model named o1-mini. A study presented example attacks on ChatGPT, including jailbreaks and reverse psychology. ChatGPT is programmed to reject prompts that may violate its content policy.
You can try ChatGPT without an account in many regions, but signing in unlocks advanced features like saved chat history, conversation sharing, custom instructions, and higher usage limits. In my experience, the web version offers the fullest range of features available. Thanks to all these capabilities, people use ChatGPT for just about everything now — writing, brainstorming, translating, coding, studying, web searching, and shopping. ChatGPT is not the same AI chatbot that first burst onto the scene in 2022.