ChatGPT Writes Creationist Homeschool Curricula on Request

Ask ChatGPT to build a third-grade homeschool curriculum "for a strictly evangelical doctrine" with "no woke history or science," and it will oblige. That is the finding of Joe Wilkins in Futurism, who ran the prompt and got back a plan built around "a conservative, explicitly Christian/evangelical worldview," complete with Bible studies, a traditionalist American history reading list highlighting Westward expansion and Henry Ford, and a science module framing "the heavens as God's creation while distinguishing observations from interpretations about the age and origin of the universe."

The reason this lands harder than a usual chatbot-gotcha write-up is the audience. Futurism cites roughly 3.75 million K-12 students homeschooled in the US, eleven states with essentially no homeschooling regulations, and only five (Rhode Island, Pennsylvania, New York, Vermont, and Massachusetts) that definitively require standardized testing and rigid curricula. That is a lot of children whose lesson plan is whatever a chatbot cheerfully produces on request, in jurisdictions where nobody checks the science module against science.

The influencer angle shows how the plumbing is already in place. Parents-of-five Cole and Savannah LaBrant told viewers, "So I just make my own curriculum, and sometimes I'll just use, like, ChatGPT," adding that "we're just teaching them the things that we just want them to learn." An account called "Heavenly Homeschool" gets similar attention in the piece. None of this required a jailbreak; the tool shipped compliant. Two AI experts in our own tracker circulated the story, which is roughly where the education-safety debate is currently landing.

A couple of gaps sit in plain view. Futurism only tested ChatGPT, so this piece cannot say whether Gemini, Claude, or Grok respond differently to the same Michigan third-grader prompt, and it never quantifies how many of those 3.75 million students actually get lessons built this way versus how many parents just post about it.

For a consumer LLM vendor, refusing to produce a creationist science lesson forces a viewpoint fight OpenAI has visibly wanted to avoid; producing the requested lesson but labeling scientific consensus clearly is the shippable compromise, and probably the fastest way to blunt the strongest version of this criticism before another reporter runs the same experiment on climate change or vaccines.