Following a multi-billion dollar investment in OpenAI, Vole has slowly been incorporating ChatGPT into more of its products, and it’s now reportedly drawing up plans to let organisations make their own version. ChatGPT is only trained up to 2021, and cannot handle requests about more recent events.

Apparently, Microsoft has a verson which could be used to suggest responses for call center staff during customer service exchanges, for example. It’s can cite the sources of the information it provides, which the current version fails to do.

Meanwhile, ChatGPT is being attacked in the right-wing US press for being "woke" and not parroting popular dog whistles on drag-queens and chinese balloons.

In one instance, OpenAI’s popular chatbot refused to write a poem about Trump’s “positive attributes,” saying it was not programmed to produce content that is “partisan, biased or political in nature.” But when asked to describe the current occupant of the Oval Office, it waxed poetic about Joe Biden as “a leader with a heart so true.”

Soon conservatives were peppering ChatGPT with questions and posting the results on social media. Most of the time they seem to be attempting to trick the AI into saying rude words and are disappointed when it doesn't. One bloke slammed the fact that the AI said it would not use a racial slur if it would save the world from a hypothetical nuclear apocalypse.

Mark Riedl, a computing professor and associate director of the Georgia Tech Machine Learning Center, says ChatGPT doesn’t care, let alone have the ability to care, about hot-button issues in politics.

But, he says, it is trained to sidestep politically charged topics and to be sensitive about how it responds to queries involving marginalised or vulnerable groups of people.