The ChatGPT moment: how one launch in November 2022 changed everything
OpenAI released ChatGPT as a quiet 'research preview' on 30 November 2022. Within two months it had an estimated 100 million users and had set off a global AI race. Here's what happened.
First published · AI Digested
On Wednesday 30 November 2022, OpenAI posted a short announcement and a link. "We've trained a model called ChatGPT which interacts in a conversational way," it said. Staff reportedly expected a modest response. Instead, they had launched one of the fastest-spreading products in history.
A "research preview"
OpenAI had been building large language models for years. Its GPT-3 model, released to developers in 2020, could write impressively, but was awkward to use and often unhelpful or offensive.
ChatGPT was built on an improved model, GPT-3.5, and trained with reinforcement learning from human feedback, where people rated its answers and the model learned to give the kind of responses people preferred. Crucially, it came with a simple chat window that anyone could use, for free.
Five days, one million users
Within five days, OpenAI's president Greg Brockman announced a million users. People shared screenshots of ChatGPT writing poems, debugging code, explaining quantum physics, drafting wedding speeches and passing practice exams.
By January 2023, analysts estimated it had reached 100 million monthly users, faster than TikTok or Instagram had done.
The race begins
The tech industry reacted fast:
- Google reportedly declared a "code red". Its chatbot, Bard, was announced in February 2023, but a factual error in its launch demo, about the James Webb Space Telescope, was followed by a fall of roughly $100 billion in the market value of Google's parent company, Alphabet, in a single day.
- Microsoft announced a multi-year, multibillion-dollar investment in OpenAI in January 2023, widely reported as about $10 billion, and put OpenAI's technology into its Bing search engine. The new Bing Chat made headlines when, in long conversations, it professed love to a journalist and argued with users.
- Meta released its Llama model to researchers in February 2023. Its weights soon leaked online, and open-weight AI took off.
- Anthropic launched Claude in March 2023.
- On 14 March 2023, OpenAI released GPT-4, which scored highly on many professional and academic exams, including a simulated bar exam.
Society reacts
ChatGPT forced AI into everyday debate.
Schools and universities scrambled. New York City's public schools blocked ChatGPT in January 2023, then reversed the ban months later. Teachers rethought homework and exams.
Workers began using it, often quietly, to draft emails, reports and code. Businesses debated whether to ban it or embrace it.
Safety warnings grew louder. In March 2023, an open letter signed by thousands, including Elon Musk, called for a six-month pause on training more powerful systems. In May, Geoffrey Hinton, one of the pioneers of deep learning, left Google to speak freely about the risks. Weeks later, hundreds of AI researchers and executives signed a one-sentence statement saying that "mitigating the risk of extinction from AI should be a global priority".
Governments moved. Italy temporarily blocked ChatGPT over privacy concerns. The UK hosted the first global AI Safety Summit at Bletchley Park in November 2023. The EU finalised its AI Act.
OpenAI's own drama. On 17 November 2023, OpenAI's board fired CEO Sam Altman, saying he had not been "consistently candid". Almost all staff threatened to quit. Within five days, Altman was back, with a new board.
Why it worked
The technology was not entirely new. What changed was access. For the first time, anyone could talk to a powerful AI model in plain language, for free. People discovered uses its creators hadn't imagined, and the world saw, all at once, how much these systems could do.
What came after
ChatGPT's launch began an era of rapid releases: models that could see images, hear and speak, reason step by step and act as agents. Hundreds of billions of dollars poured into AI chips and data centres. OpenAI said in October 2025 that ChatGPT had 800 million weekly users.
Three years on, it's hard to remember that AI chatbots were a niche curiosity until a Wednesday in late 2022.
Read on
Common questions
Was ChatGPT the first chatbot?
No. Chatbots date back to ELIZA in the 1960s, and the model behind ChatGPT, GPT-3.5, was similar to ones developers already had access to. What was new was a free, easy chat interface and training that made it a helpful conversational assistant.
How many people use ChatGPT?
OpenAI said in October 2025 that ChatGPT had 800 million weekly users. Numbers have continued to change since; see our news section for the latest.
Why was it called ChatGPT?
'Chat' because you talk to it, and 'GPT' after the generative pre-trained transformer model underneath. OpenAI has said the name was chosen in a hurry before launch.
Sources and further reading
Spotted a mistake or something out of date? Tell us and we will fix it. Read how we research and write our articles.
Read next
How ChatGPT works: large language models explained simply
How do chatbots like ChatGPT, Claude and Gemini actually work? A plain-English explanation of large language models: tokens, training, prediction and why they sometimes get things wrong.
Updated 22 November 2026
ChatGPT, Claude, Gemini or Copilot? Choosing an AI assistant
An honest, plain-English comparison of the main AI assistants: what each is best at, what's free, what the paid plans add, and how to choose the right one for you.
Updated 22 November 2026
How to use ChatGPT and other AI chatbots: a complete beginner's guide
Never used an AI chatbot? This step-by-step guide shows you how to sign up for free, ask your first questions, get better answers, and stay safe while you do it.
Updated 22 November 2026