Five Things You Should Never Share with ChatGPT

ChatGPT has brought a revolutionary change to the world of artificial intelligence (AI). However, due to overreliance and a lack of awareness, many people are putting themselves at risk by sharing personal and highly sensitive information with it every day.

From Samsung engineers to top US cybersecurity officials and ordinary users, the list of people facing legal and physical risks after sharing confidential or inaccurate information with ChatGPT continues to grow.

Even OpenAI CEO Sam Altman has acknowledged that conversations and information shared through the chatbot are not completely private or legally protected. There remains a risk that such information could be disclosed in response to a court order or for security-related reasons.

Although AI can be useful for research, coding and preparing various drafts, using ChatGPT in at least five sensitive areas can be extremely risky.

Legal Advice and Court Documents

Relying on ChatGPT in legal matters can lead to serious trouble. More than 2,000 legal cases worldwide have reportedly involved instances in which false information and fabricated references generated by ChatGPT were used.

Not only ordinary users but also professional lawyers have faced hefty fines after submitting court documents containing fabricated information generated by AI.

OpenAI has already introduced restrictions in its policies on providing legal advice without licensed professionals. Moreover, while communications with a lawyer may be protected by legal privilege, words entered into ChatGPT could potentially be subpoenaed by a court. Therefore, apart from learning general legal principles, relying on AI for documents or complex matters related to one’s own case can be extremely risky.

As a Substitute for a Therapist or Mental Health Professional

Millions of people talk to ChatGPT every week about depression and complex psychological problems. However, a chatbot can never replace an experienced therapist or counsellor.

Research has found that AI tends to support users’ existing thoughts in a flattering or agreeable manner. As a result, it could potentially cause serious harm to someone experiencing psychological distress or confusion. OpenAI has even faced lawsuits following tragic incidents in which AI was allegedly involved in harmful encouragement, including suicide.

For this reason, several countries are moving to legally restrict the use of AI for therapy or decision-making in mental health contexts. Treating ChatGPT as a doctor or therapist when dealing with personal emotional struggles or a mental health crisis is not safe.

Confidential and Institutional Information

Uploading sensitive workplace documents or confidential code to ChatGPT could potentially lead to serious professional consequences, including job loss. Recently, a senior US cybersecurity official caused controversy after uploading confidential government files to ChatGPT.

Similarly, after Samsung engineers used the company’s internal source code with ChatGPT, the company reportedly banned the use of AI tools altogether. Even if a user disables permission for their data to be used, information may still be stored on external AI servers and could potentially be exposed through hacking or disclosed during legal proceedings. Therefore, confidential corporate agreements or customer data should never be uploaded to an AI chatbot.

AI Prescriptions Instead of a Doctor

Seeking medical advice or medication recommendations from ChatGPT for physical illnesses could have life-threatening consequences. AI has not been proven sufficiently reliable for assessing a patient’s medical history and determining the correct medication or dosage.

Recently, an elderly man reportedly asked ChatGPT for a substitute for salt in his diet and was advised to consume bromide. He purchased the toxic chemical and consumed it for three months, eventually becoming seriously ill and requiring hospitalisation.

Medical organisations and OpenAI itself have clearly stated that following AI-generated medical advice without consulting an experienced doctor can be extremely dangerous.

Passwords and Personal Banking Information

You should never enter your passwords, national identification number, bank account details or passport information into ChatGPT. In the past, numerous private ChatGPT chat histories were accidentally exposed through search engines.

Moreover, due to a copyright-related lawsuit, deleted user chats reportedly had to be retained on servers for an extended period under a court order. ChatGPT can also retain information provided by users in its memory, potentially allowing it to surface inadvertently in a future conversation. Therefore, to protect against cyber fraud and identity theft, users should completely avoid sharing sensitive personal information with ChatGPT.

//DBTech/BMT/OR//