Artificial intelligence is not your doctor, be careful when consulting

Chatbots are changing the way we think, learn and perceive the world around us. This type of transformation is evident in many areas of life, but perhaps one of the most sensitive and often worrying is the growing use of generative artificial intelligence (GenAI) tools for healthcare. ESET, a world leader in cybersecurity, has warned that using artificial intelligence to obtain medical advice may expose it to security and privacy risks. He listed the risks and shared his recommendations for how to stay safe.

Along with a number of free AI chatbots, major tech companies have entered the consumer health AI space with the launch of services like Copilot Health, ChatGPT Health, and Amazon's HealthAI, which help users interpret their medical records and ask questions about their symptoms, lab results, and treatment options. Expecting an AI tool to take over the role of a doctor poses some risks. The risk is not only that users receive the wrong advice, but that they share highly sensitive personal information with systems whose privacy protections, data sharing practices and legal obligations may differ from those of a doctor or hospital, and that their data is exposed to unexpected organisations. Misuse of AI chatbots in general is now the number one threat in healthcare technology, according to a US patient safety organization.

People talk about their own health and the health of their loved ones on their mobile devices more than any other topic, according to Microsoft. Chatbots are available 24/7 and speak in a confident tone that helps put nervous patients at ease while offering an answer for everything. At a time when national healthcare systems are increasingly under pressure, many people will likely self-diagnose with the help of AI before deciding to seek medical help.  However, some concerns have already begun to emerge. The first of these is hallucinations or false advice. An Oxford University study published in February in the journal Nature Medicine found that: Users often do not know what information they should share with LLM;  LLMs give very different answers even when the questions posed to them vary slightly, and  models often give both good and bad advice, but users have a hard time distinguishing between the two. According to the chief physician of the study, despite all this excitement, artificial intelligence is not yet ready to take on the role of a physician. Patients should be aware that consulting a large language model about their symptoms can be dangerous, resulting in misdiagnoses and not recognizing when immediate help is needed. 

Artificial intelligence privacy risks

There are also non-health-related risks that will require individuals to stop and think for a moment. Sharing sensitive medical information with a public chatbot could mean that the data is used to train the model and therefore passed on to others. Models have been known to accidentally disclose data entered by their users, albeit unintentionally. Your health data is not like a stolen credit card that can be frozen while its information is changed and rearranged. It's yours for life, and when shared with an AI tool, it can become a permanent digital record. Most of the major health-focused chatbots promise not to use this data for educational purposes. Education is just one piece of the privacy picture, and services may not make the same promises when it comes to third-party data sharing. Your personal medical information may be obtained by a data collector, a third party between the model provider and your healthcare provider. It may also be shared with advertisers directly or through one of these aggregators, although it is usually anonymized before use. Health data is extraordinarily sensitive and anonymization does not always eliminate all risks.

When the risk of breach multiplies

Health data can be highly monetized by fraudsters for a variety of reasons. It retains its value for a long time because it generally cannot be changed or rearranged.It may contain insurance information that can be used to submit fraudulent claims or obtain medical services on your behalf. It can be used to blackmail you. The more companies hold this data, the more likely hackers are to seize and steal this data. 

Be careful what you share with AI

If you're concerned about a health issue, stay away from general-purpose bots and instead choose bots specifically designed to answer health-related questions. Check whether the service discloses how it processes your data, uses commands you enter for training purposes, shares information with third parties, and is covered by a privacy regulation. Don't blindly trust the output if there are no source links to verify. Do not accept answers as absolute truth: Always consult a medical professional or an official website.

To protect your privacy, consider the following:

• Never share or upload medical documents, lab results, or other sensitive documents with an AI tool unless you understand how the tools process that data.
• Avoid entering name, address, insurance information, patient numbers, or other identifying information. avoid.
• Make sure education and chat history features are turned off.
• Share the minimum information necessary for the task.
• Assume that anything you type may be withheld or disclosed, and adjust your commands accordingly.

AI chatbots can be useful for brainstorming questions to ask your doctor about a particular condition or explaining a medical term you're unfamiliar with. But there is a big difference between using AI to prepare for medical care and using AI to replace medical care. Don't accept a confident answer as a diagnosis and don't ignore immediate symptoms just because a machine puts you at ease.