Claims have emerged that Google developed an AI chatbot a year before OpenAI but refrained from releasing it due to concerns it could threaten its existing search business. The news surfaced after an AI engineer posted on X on August 1, writing about an interview where Jeff Dean mentioned having an in-house bot before ChatGPT. Thibault Sotty, former head of Codex at Google and DeepMind who joined OpenAI in 2024, responded, “I was part of that team. It was essentially ChatGPT created a year before ChatGPT.” He added that the chatbot was initially called ‘LMChat’ and that DeepMind was blocked from releasing a product that could disrupt Google. Jeff Dean confirmed in a podcast interview last year that Google had a chatbot named “Meena” before ChatGPT’s release, used by employees during the COVID-19 pandemic. However, Google evaluated the chatbot against its search engine standards, and Meena suffered from “hallucinations” and aggressive responses. Dean acknowledged Google failed to recognize the AI’s potential in tasks like drafting documents or summarizing lengthy texts. Google had pioneered deep learning, acquiring DeepMind in 2014 and introducing the “Transformer” architecture in 2017. Former Google researchers Daniel de Freitas and Noam Shazeer left in 2021 after repeated rejections for their chatbot proposal, later founding Character.AI. Google’s top-tier model “Gemini 3.5 Pro,” initially slated for a June release, was postponed due to