Elon Musk is reportedly considering a new AI research lab focused on an alternative to ChatGPT. The project is at an early stage: there are no concrete product plans, and Igor Babuschkin, a scientist who recently worked at Google’s AI unit Deepmind, has not signed a contract.
A reported project without a product plan
The Information reported that Musk aims to launch the lab. Babuschkin, who previously worked at OpenAI, is expected to take a key position. OpenAI was co-funded by Musk, who was one of its main funders.
Musk left OpenAI’s board in 2018 to avoid conflicts of interest with Tesla, his electric car company, which is also researching AI. The report describes a possible return to the field through a separate effort, though the details remain unsettled.
Babuschkin said he would like to “work with Elon on something in the LLM space.” That statement signals interest, not a confirmed appointment or a defined product. Without a signed contract or concrete plans, the reported lab remains an idea in development.
Concerns about ChatGPT’s answers
Musk has voiced sharply different reactions to ChatGPT. He once called its progress “scary good” and said, “We are not far from dangerously strong AI.” Later, he described some of its answers as “Extremely concerning.”
Those concerns followed reports about the chatbot refusing to define a woman, praising Democrats instead of Republicans, and claiming nuclear weapons were less dangerous than racism. Musk also said the chatbot was too “woke” for him, based on interactions on Twitter.
Babuschkin said the proposed chat AI should argue better and be more objective. He also described the goal as making the model more trustworthy and reliable. The source says the aim is not simply to give the model fewer rules.
What “TruthGPT” suggests
In mid-February, Musk called for “TruthGPT.” The name points to his interest in a model that tells “the truth” from his perspective, rather than one shaped around certain social expectations.
That framing connects the proposed project to Musk’s stated desire to strengthen what he sees as free speech on Twitter. The article suggests that a more politically neutral ChatGPT alternative would fit that aim, while making clear this is an inference about his possible intentions rather than a confirmed product description.
Words such as “truth” and “objectivity” can describe different expectations for a chatbot. The reported project’s specific approach to those ideas is not yet clear. No technical details or product commitments are provided.
Customization may shape the competition
OpenAI has announced plans for customizable ChatGPT models. Users will be able to tailor future models to their needs, within limits set by society, according to the company.
Separately, leaked information about OpenAI’s Foundry development platform mentions “more robust fine-tuning options for the latest models.” If models become more customizable, users may be able to adjust how they respond, which could help address some criticism about political bias.
For now, the reported Musk project has no concrete product plans, while OpenAI has discussed future customization. Whether either path will satisfy users seeking chat AI that feels more balanced, reliable, or suited to their preferences remains unresolved.