O1 is designed to solve more complex problems by spending more time “thinking” before it answers, enabling it to analyze its answers and explore different strategies. In September 2024, OpenAI introduced o1-preview and a faster, cheaper model named o1-mini. On July 18, 2024, OpenAI released GPT-4o mini, a smaller version of GPT-4o which replaced GPT-3.5 Turbo on the ChatGPT interface. OpenAI has not revealed technical details and statistics about GPT-4, such as the precise size of the model. The term “hallucination” as applied to LLMs is distinct from its meaning in psychology, and the phenomenon in chatbots is more similar to confabulation or bullshitting. A 2023 analysis estimated that ChatGPT hallucinates around 3% of the time.
In September 2025, OpenAI added a feature called Pulse, which generates a daily analysis of a user’s chats and connected apps such as Gmail and Google Calendar. In December 2024, OpenAI launched a new feature allowing users to call ChatGPT with a telephone for up to 15 minutes per month for free. An optional “Memory” feature allows users to tell ChatGPT to memorize specific information. ChatGPT’s training data includes software manual pages, information about internet phenomena such as bulletin board systems, multiple programming languages, and the text of Wikipedia.
These limitations may be revealed when ChatGPT responds to prompts including descriptors of people. ChatGPT’s training data only covers a period up to the cut-off date, so it lacks knowledge of recent events. To implement the feature, https://www.luckytwicecasino.eu/ OpenAI partnered with data connectivity infrastructure company b.well. On January 7, 2026, OpenAI introduced a feature called “ChatGPT Health”, whereby ChatGPT can discuss the user’s health in a way that is separate from other chats. In 2025, OpenAI added several features to make ChatGPT more agentic (capable of autonomously performing longer tasks).
The official app by OpenAI
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 July 2025, OpenAI released ChatGPT agent, an AI agent that can perform multi-step tasks. It was capable of autonomously performing tasks through web browser interactions, including filling forms, placing online orders, scheduling appointments, and other browser-based tasks. The integration used ChatGPT to write prompts for DALL-E guided by conversations with users. At launch, OpenAI included more than 3 million GPTs created by GPT Builder users in the GPT Store.
- ChatGPT (based on GPT-4) was better able to translate Japanese to English when compared to Bing, Bard, and DeepL Translator in 2023.
- ChatGPT’s training data only covers a period up to the cut-off date, so it lacks knowledge of recent events.
- The user can interrupt tasks or provide additional instructions as needed.
- ChatGPT has been used to generate introductory sections and abstracts for scientific articles.
- To implement the feature, OpenAI partnered with data connectivity infrastructure company b.well.
- If you ask directly, the model will give you competent, but not very inspiring, buying advice.
- These rankings were used to create “reward models” that were used to fine-tune the model further by using several iterations of proximal policy optimization.
Data Linked to You
ChatGPT has been used to generate introductory sections and abstracts for scientific articles. Additionally, using a model’s outputs might violate copyright, and the model creator could be accused of vicarious liability and held responsible for that copyright infringement. When assembling training data, the sourcing of copyrighted works may infringe on the copyright holder’s exclusive right to control reproduction, unless covered by exceptions in relevant copyright laws. Juergen Schmidhuber said that in 95% of cases, AI research is about making “human lives longer and healthier and easier.” He added that while AI can be used by bad actors, it “can also be used against the bad actors”.
Model versions
ChatGPT is a chatbot and AI assistant built on large language model (LLM) technology. OpenAI collects data from ChatGPT users to further train and fine-tune its services. These rankings were used to create “reward models” that were used to fine-tune the model further by using several iterations of proximal policy optimization. ChatGPT is based on GPT foundation models that have been fine-tuned for conversational assistance. It uses generative pre-trained transformers (GPTs), such as GPT-5, to generate text, speech, and images in response to user prompts.
- ChatGPT was initially free to the public and remains free in a limited capacity.
- Kelsey Piper of Vox wrote that “ChatGPT is the general public’s first hands-on introduction to how powerful modern AI has gotten” and that ChatGPT is “smart enough to be useful despite its flaws”.
- An optional “Memory” feature allows users to tell ChatGPT to memorize specific information.
- However, no machine translation services match human expert performance.
- In the case of supervised learning, the trainers acted as both the user and the AI assistant.
Data safety
OpenAI’s outsourcing partner was Sama, a training-data company based in San Francisco, California. These labels were used to train a model to detect such content in the future. The ethics of its development, particularly the use of copyrighted content as training data, have also drawn controversy. The chatbot can facilitate academic dishonesty, generate misinformation, and create malicious code. The service gained 100 million users in two months, making it the fastest-growing consumer software application in history.
In the UK, a judge expressed concern about self-representing litigants wasting time by submitting documents containing significant hallucinations. In November 2025, 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 medical education, it can explain concepts, generate case scenarios, and be used by students preparing for licensing examinations. 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.
Their leaders emphasized their earlier caution regarding public deployment was due to the trust the public places in Google Search. Kelsey Piper of Vox wrote that “ChatGPT is the general public’s first hands-on introduction to how powerful modern AI has gotten” and that ChatGPT is “smart enough to be useful despite its flaws”. As before, OpenAI has not disclosed technical details such as the exact number of parameters or the composition of its training dataset.
Google Drive
Shortly after the bug was fixed, users could not see their conversation history. In March 2023, a bug allowed some users to see the titles of other users’ conversations. Despite this, users may jailbreak ChatGPT with prompt engineering techniques to bypass these restrictions.
The chatbot can assist patients seeking clarification about their health. The uses and potential of ChatGPT in health care has been the topic of scientific publications and experts have shared many opinions. Many companies adopted ChatGPT and similar chatbot technologies into their product offers. The Guardian questioned whether any content found on the Internet after ChatGPT’s release “can be truly trusted” and called for government regulation. In June 2023, hundreds of people attended a “ChatGPT-powered church service” at St. Paul’s Church in Fürth, Germany.
The model’s baseline tone is practical, but a perspective shift convinces it to behave like a storyteller instead of a rulebook. If you ask directly, the model will give you competent, but not very inspiring, buying advice. I want a hot take that sounds like it would start a family argument.” The trick here is to acknowledge what the model typically gives you and frame the request as a deviation. ” The tasks are still useful, but the tone shifts from bland to bracing.
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. Robin Bauwens, an assistant professor at Tilburg University, found that a ChatGPT-generated peer review report on his article mentioned nonexistent studies. 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”. 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”. In July 2023, the FTC launched an investigation into OpenAI over allegations that the company scraped public data and published false and defamatory information.