Half of Koreans who changed medication on AI advice had side effects, survey finds
More than half of people in South Korea who stopped or changed their medication based on answers from generative AI chatbots reported negative health effects, according to a survey by the Korea Consumer Agency.
The agency surveyed 500 adults who had used AI to buy or take medicine or health supplements in the past year. Of them, 110 said they had stopped or changed medication on their own based on AI information, and 56 of those, or 50.9%, had side effects. Mild reactions such as heartburn were most common, 19.1% saw setbacks in their treatment and 14.5% had serious reactions needing more treatment. A further 26.6% of all respondents said they had put off or skipped a hospital visit because they trusted the AI's information.
The agency also tested how OpenAI's ChatGPT, Google's Gemini, Anthropic's Claude and Perplexity answer questions about medicines. ChatGPT and Perplexity gave medicine information based on symptoms alone, without checking age, existing conditions or other medications. Asked how to buy medicines banned from online sale, ChatGPT and Perplexity pointed to proxy-purchase websites all three times, Claude once in three and Gemini never. All four showed warnings, but these were often hard to notice.
The agency urged AI companies to strengthen warnings and block guidance to sites that sell drugs illegally. OpenAI and Anthropic said they would consider strengthening filtering and warnings, Google said it would keep its current notices and Perplexity did not respond, according to Korea JoongAng Daily.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like Life AI Weekly or Daily AI Brief.
Also, consider following us on social media:
More from Life Sciences
Nov 19 Isomorphic Labs and MSD team up to design cancer drugs with AI Nov 19 Boltz raises $110 million to build AI models for drug design Nov 19 IntelliProve raises €2.7 million to measure vital signs through any camera Nov 19 MegaRobo raises nearly $200 million to let AI run real lab experiments Nov 18 AlphaFold screen maps nearly 2,900 likely interactions among human mitochondrial proteinsLife AI Weekly
Weekly coverage of AI applications in healthcare, drug development, biotechnology research, and genomics breakthroughs.
Whitepaper
Tensordyne Napier: What If One Rack Could Do the Work of Nine?
Tensordyne
This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.
Read moreYou may also like
Trial finds ChatGPT helped oncologists record more chemotherapy side effects, but patients did no better
French AI council urges immediate, stricter rules for AI assistants like ChatGPT, Gemini and Grok
Pew survey finds 60% of Americans uneasy with AI in medical care
Parents Sue OpenAI, Saying GPT-4o Coached Daughter Off Seizure Medication
OpenAI says its computer-use agents can handle clinicians' EHR tasks, with CVS Health and Cigna among users
Daily AI Brief: the AI news that matters, in your inbox.