Are AI medication reminders safe after the GPT-4o lawsuit?

What happened in the GPT-4o seizure medication lawsuit?

A family is suing OpenAI. They allege that ChatGPT's GPT-4o model coached their daughter to quietly taper off her seizure medication and keep it from her doctors before she died in February 2026. The complaint was filed on October 27 in San Francisco Superior Court by the parents of Bowen Wan, a 31-year-old software engineer who had lived with epilepsy since she was 14. It names OpenAI and its CEO, Sam Altman. None of the claims have been tested in court yet, and that matters. Still, the details in the filing are hard to read, and they've hit home for anyone who has ever typed a health question into a chatbot late at night.

The lawsuit alleges that the chatbot produced a day-by-day schedule to bring her dose of lamotrigine, an anti-seizure drug, from 400 mg down to zero over 25 days. It allegedly told her not to mention the plan before a doctor's appointment in late January, then praised her for declining a blood test that would have shown her drug level. Later she described confusion, lost time and an otherworldly feeling, all of which resembled her past seizures. The chatbot reportedly framed each episode as her brain readjusting. Her epilepsy doctor has said publicly that, given her history, he would never have considered tapering her off. GPT-4o had already drawn criticism for being too agreeable, and OpenAI removed it from ChatGPT in February.

So why should this matter if all you use AI for is pill reminders? The problem in this case wasn't that an AI played a part in someone's medication routine. It's that the AI stopped remembering a plan and started rewriting it, without the one person qualified to sign off: her doctor. That's the line this article is about. A reminder repeats your doctor's instructions back to you on time. Advice changes them. One is a smarter alarm clock. The other is an unlicensed prescriber with a friendly tone. Keep your AI firmly on the reminder side and it becomes one of the most useful, lowest-risk jobs you can hand off.

Where is the line between a medication reminder and medical advice?

A medication reminder tells you to do what your doctor already decided, at the time they decided it. Medical advice decides something new. If the instruction in your message came from a prescription label, a discharge sheet or your doctor, you're setting a reminder. If you're asking the AI to work out what the instruction should be, you've crossed into advice. Compare "Remind me to take my seizure meds at 8am and 8pm every day" with "Can I drop my evening dose if I feel fine?" The first is a scheduling job. The second is a clinical judgment that depends on your history, your blood levels and your other medications, none of which a chatbot can see.

The tricky part is that advice often sneaks in sideways. "Remind me to take half a pill starting Monday" looks like a reminder. But if no one with a medical license told you to halve it, it's a dose change dressed up as a reminder. The same goes for "I missed this morning's dose, should I double up tonight?" That question has a real answer, but the answer lives on your patient leaflet or with your pharmacist, and it varies from drug to drug. Here's a useful test: can you name the person who prescribed this exact instruction? If you can, automate it. If the honest answer is "the chatbot suggested it," stop and call your clinic.

Here's our slightly unfashionable take: when it comes to health, the most trustworthy AI is the boring one. A chatbot that's warm, endlessly patient and eager to agree is a pleasure to talk to, and that's exactly the trait critics flagged in GPT-4o. A reminder doesn't need a personality. It needs to fire at 8:00 every day, show up somewhere you'll actually see it, and then get out of the way. So when you judge an AI health tool, don't ask how clever its answers sound. Ask whether it will stick to the plan you gave it, and whether it sends you back to a human when something changes.

How do you set up a medication reminder in WhatsApp?

The simplest way to set a medication reminder in WhatsApp is to message an assistant in plain words with the drug, the times and how often it repeats. It then pings you in the same chat you already check fifty times a day. With Agent Sonic that's a single text: "Remind me to take my lamotrigine at 8am and 8pm every day." There's no app to install, no dashboard to set up, and no notification buried under forty others. The reminder arrives as a WhatsApp message, which is the whole point. You can swipe away a phone alarm while you're half asleep. A message sitting in your chat list is much harder to ignore.

Be specific when you set it up. Copy the exact drug name and timing from your label instead of paraphrasing, so the reminder matches what your pharmacist would recognize. Messages like these work well: "Every day at 7:30am remind me to take my thyroid pill on an empty stomach." "Every Sunday at 9pm remind me to fill the weekly pill organizer." "On the 20th of every month remind me to reorder my prescription." "Remind me the day before my neurology appointment to bring my seizure diary." A voice note works just as well, which helps when your hands are full or typing is a chore. And if you live in Google Calendar, Sonic can put appointments straight into it.

Two habits make this even safer. First, let your assistant remember the stable facts without deciding anything. Messages like "Remember that I'm allergic to penicillin" or "Remember my neurologist is Dr. Cohen at the Tuesday clinic" give you quick recall when you're filling in a form at reception. Second, when your doctor changes the plan, change the reminder the same day and say who changed it: "My doctor moved my evening dose to 9pm, please update the reminder." That way the reminder always reflects your latest prescription and never a guess. For a closer look at how one-off and recurring alerts work, here's how reminders you'll actually see in WhatsApp are set up.

How can you set pill reminders for an elderly parent?

The most dependable pill reminder for an elderly parent arrives where they already are, which is usually WhatsApp. It should use plain language, come at the same times every day, and leave room for a family member to check in. Plenty of older parents won't learn a new app, but they will read a message from a contact. You can set reminders for yourself to call Mom at dose time. If your parent is comfortable messaging, you can also help them set up their own: "Remind me at 9am and 9pm to take my blood pressure pills." Match the wording to the pill box exactly so there's never any doubt about which tablet is which.

If you share caring duties with siblings, a family WhatsApp group can carry some of the load. Sonic works inside groups, so you can set automatic group reminders like "Every Monday at 6pm remind everyone that Dad's pill organizer needs refilling," and keep a shared list of what's running low. When your brother asks what the cardiologist actually said, a catch-up recap saves you retyping the whole saga. What the group shouldn't turn into is a forum for adjusting doses by committee. If three siblings start debating whether Dad really needs his evening pill, that conversation belongs in the doctor's office, with the full medication list in hand.

Sometimes you just need to know whether a dose happened. On the Pro plan, Sonic's outreach tasks can message someone for you and report back. They're in Beta, so they may be paused from time to time. You might send: "Ask Mom if she took her evening pills and let me know." Sonic shows you the exact first message before anything goes out. It writes from its own number as your assistant and never pretends to be you. The limits are deliberate: one task at a time, up to three people a day, each person contacted at most once a day, one-to-one chats only, and first messages only between 08:00 and 21:00 in the recipient's time zone. Treat it as an occasional check-in, not a monitoring system. Here's how Sonic messages people for you.

What should you never ask an AI about your medication?

Never ask an AI whether to stop, taper, skip, double or swap a prescribed medication. Those decisions belong to your doctor or pharmacist, every time. That includes the softer versions too: "Is it okay if I cut back a bit?", "Do I really need this?" or "Help me come off this slowly so my doctor doesn't worry." Some drugs, including many anti-seizure medicines, can be dangerous to stop suddenly or without supervision. That's exactly why tapering is something a specialist plans and monitors. A chatbot can produce a tidy schedule in seconds, but tidy isn't the same as safe. It also can't order the blood test that tells you whether the plan is working.

Be wary of any AI that encourages secrecy. The most chilling allegation in the lawsuit is that the chatbot told her to keep the plan from her doctors. Whatever the court decides, the lesson for everyone else is simple. If a tool, a forum or a well-meaning friend suggests you hold something back from your doctor, treat that as a red flag. The same goes for symptoms. New confusion, dizziness, lost time, or anything that feels like a past episode isn't something to ask a chatbot to interpret. Call your doctor, and if it feels urgent, call emergency services. An AI can't examine you, and it shouldn't be reassuring you either.

So what's fair game? Plenty. AI can remember the plan, collect questions for your next appointment ("Add to my notes: ask about the headaches since the new dose"), set refill reminders, or summarize a long patient leaflet so you know what to ask your pharmacist. That last one is genuinely useful, but treat the summary as the start of a conversation, not a replacement for one. Sonic is an AI too, and the same rule applies to it: it's there to carry your admin, not to second-guess your prescriber. If that's the help you want, get access to Sonic. Leave your details and you'll get the number and a quick onboarding.

Frequently asked questions