OpenAI Executive Says Google Had ChatGPT Before OpenAI - Chose Not to Release It

Nearly four years after ChatGPT changed how the world interacts with technology, a fresh claim is making the rounds about what Google was sitting on before that moment. Thibault "Tibo" Sottiaux, who now works at OpenAI but previously spent time on Google's AI team, says Google had built something remarkably close to ChatGPT well before OpenAI's chatbot ever launched - the company simply chose not to put it out into the world.

The claim surfaced when Tibo responded to a post on X from AI engineer Cheng Lou, who wrote, "I sometimes think about that Jeff Dean interview where he said they had an internal bot before ChatGPT but didn't think it was better than just googling."

Tibo, who was part of that internal effort at Google before eventually joining OpenAI, confirmed the project was real and further along than many might assume. "I was part of that team. Basically ChatGPT one year before it came out. Called LMChat and then another codename," he wrote.

Google Reportedly Held Back Over Search Concerns

According to Tibo, the decision to shelve the chatbot wasn't about capability - it was about risk. He suggested Google feared that releasing such a product could threaten its core search business, and that DeepMind faced internal restrictions preventing it from shipping anything that might disrupt that business. "Google was too nervous to release it and DeepMind was blocked from shipping products that could disrupt Google," Tibo added. "I think about this a lot."

Tibo left Google DeepMind in July 2024 to join OpenAI

It's worth remembering that the very foundation of today's large language models traces back to Google itself, through its landmark 2017 research paper "Attention is All You Need," which introduced the Transformer architecture - the same "T" that appears in GPT, or Generative Pre-trained Transformer. Despite laying that groundwork, Google opted against releasing its own chatbot at the time, leaving the door open for OpenAI to define the moment instead. Google later found itself catching up, first with Bard and eventually with Gemini.

Back in 2022, Google CEO Sundar Pichai and DeepMind chief scientist Jeff Dean acknowledged that Google possessed similar technology to what OpenAI had built, but argued that the stakes were higher for Google since users place a high degree of trust in the accuracy of its answers. Dean reportedly told employees during an all-hands meeting, "This really strikes a need that people seem to have but it's also important to realize these models have certain types of issues," pointing to concerns around hallucinations - a challenge that still affects AI systems today.

Prior to ChatGPT's debut, Google was widely regarded as the frontrunner in AI research, backed by the combined strength of Google Brain, DeepMind, and Google Research, along with earlier systems like LaMDA, PaLM, and Meena. Much of that work reportedly stayed internal, held back by worries over accuracy, safety, and the potential fallout for search.

Once ChatGPT arrived, the pace of the industry shifted dramatically. The chatbot hit one million users within five days and 100 million within two months. Today, ChatGPT counts roughly 1 billion monthly active users, with Gemini close behind at nearly 950 million, as Google, OpenAI, and Anthropic continue to compete at the front line of AI development.