ChatGPT adds a study mode that refuses to do your homework

Most students use ChatGPT like an unpaid intern: drop in the homework prompt, copy the answer, and close the tab. But that habit comes with receipts. A study released in June found that people using ChatGPT to write essays showed lower brain activity than those using Google Search or their own brains.

Now OpenAI is trying to fix the brain-rot dilemma. On Tuesday, the company launched Study Mode in ChatGPT, a dedicated tool that refuses to just hand over the answers and instead acts like a persistent tutor.

The feature is available now for logged-in users on Free, Plus, Pro, and Team plans, with ChatGPT Edu rolling out over the coming weeks.

The Socratic pivot

To turn it on, users select "Study and learn" from the tools menu. Once active, the model changes its personality.

Instead of spitting out a completed calculus problem or an essay outline, ChatGPT uses scaffolded responses and Socratic questioning. It nudges you with hints, breaks dense subjects into bite-sized chunks, and asks follow-up questions to gauge whether you actually grasp the material. In some cases, it outright refuses to give you the final answer until you show your work.

OpenAI built the pedagogical guidelines alongside educators and learning scientists. Interestingly, the feature currently runs on custom system instructions rather than a dedicated fine-tuned model. OpenAI admitted this might lead to occasional inconsistencies, but said it plans to bake these interactive behaviors directly into future foundation models once it gathers enough user data.

The loophole: You can just turn it off

Anthropic beat OpenAI to this punch back in April with its own Learning Mode for Claude. But both companies share the exact same Achilles' heel: self-control.

Study Mode is completely optional. If a student is staring at an 11:59 PM deadline, they can toggle Study Mode off mid-chat and get back to copy-pasting.

Leah Belsky, OpenAI’s VP of Education, confirmed to TechCrunch that the company currently does not offer controls for parents or school administrators to lock students into Study Mode, though they might explore administrative controls down the road.

OpenAI is also partnering with Stanford University's SCALE Initiative through its NextGenAI project to research how these conversational dynamics affect learning outcomes in K-12 settings.

Why it matters

For independent learners and college students tackling genuinely tough topics, this is a massive upgrade. You no longer have to spend five minutes prompt-engineering ChatGPT to "act like a professor and quiz me." It works out of the box as a patient, 24/7 sounding board.

For OpenAI's enterprise strategy, this is defense turned into offense. After years of school districts banning and unbanning ChatGPT over plagiarism and cognitive decline fears, OpenAI desperately needs to position itself as a net-positive educational tool. Making this pitch is essential if it wants school boards to approve massive enterprise deals for ChatGPT Edu.

We think launching this via prompt scaffolding first is a smart, low-risk way to stress-test pedagogical workflows. But the elephant in the room is enforcement. Expecting teenagers to voluntarily choose Socratic struggle over an instant answer is wishful thinking. Until schools and parents get administrative locks, Study Mode will likely remain a tool for the highly motivated minority rather than a fix for the broader classroom crisis.