Artificial intelligence retains 30 days the medical or banking data that users have just deleted
Artificial intelligence retains 30 days the medical or banking data that users have just deleted
Google News - AIArtificial intelligence retains 30 days the medical or banking data that users have just deleted
Experts advise against using AI for medical or financial consultations. Although OpenAI allows disabling training, sensitive data requires anonymization and deleted chats take 30 days to be removed.
Add APD as a preferred Google source for free.
Stay informed with the latest current news.
Cybersecurity experts ask users not to enter personal, medical, financial, or work data into artificial intelligence chatbots, even when used for everyday queries.
The warning clashes with one of the most widespread uses of these tools, as many people turn to them precisely to review documents, resolve health doubts, or ask for help with budgets, three areas in which the exposure of sensitive information multiplies the risk.
OpenAI allows disabling the use of new chats to train the model
In ChatGPT, users can disable the option "Improve the model for everyone so that new conversations are not used for training", within the data controls section. OpenAI also explains that this setting applies to new conversations and that the temporary chat is outside the history. ([help.openai.com](https://help.openai.com/en/articles/8983130-how-does-chatgpt-use-my-data?utm_source=openai))
Deleting a message is not enough.
Deleted chats disappear from view immediately, but OpenAI indicates that their permanent deletion from systems can take up to 30 days, with legal or security exceptions. In the case of temporary chat, the company points out that it also does not appear in the history and that it keeps a copy for a maximum of 30 days for security reasons. ([help.openai.com](https://help.openai.com/es-es/articles/8809935-how-chat-retention-works-in-chatgpt?utm_source=openai))
Medical and banking documents require anonymization before uploading
The most repeated recommendation affects identification, health, and financial documents. Passports, Social Security numbers, account numbers, credit cards, or clinical reports should not be entered into a chatbot if they contain real personal data.
When it is essential to analyze a document with AI, the guideline is to hide personal data before uploading it. This implies replacing names with initials, removing identification numbers, addresses, and file numbers, and using fictitious data when the query does not require real information.
In the medical field, prior anonymization also extends to symptoms, reports, and any reference that could identify the patient. The same logic applies to financial consultations, where it is preferable to present budgets with invented amounts rather than copying authentic bank data.
Companies must review what information their employees share
The risk is not limited to domestic use and extends to work environments, where employees must avoid sharing non-public contracts, business strategies, trade secrets, or proprietary code without first checking their company's security policies.
That caution also includes credentials and access keys. Passwords, authentication codes, private keys, and access tokens must be managed with specific protection tools, not within a conversation with an automated assistant.
Furthermore, OpenAI states that it applies systems to identify and hide personal information in the datasets used for training its models, a framework that coexists with broader recommendations on privacy in the daily use of AI, as already reported by warnings about AI queries and vulnerabilities of agentic AI. ([apdnoticies.com](https://www.apdnoticies.com/es/proteccion-datos/cada-consulta-a-la-ia-entrena-sus-algoritmos-antes-de-exponer-tus-datos-a-la-venta-en-el-mercado-negro-de-catalunya_12719_102.html?utm_source=openai))
OpenAI indicates that temporary chat conversations are deleted from its systems after 30 days, except for cases related to security and abuse prevention. ([help.openai.com](https://help.openai.com/en/articles/8914046-temporary-chat-faq?utm_source=openai))
Related Stories
AI News
Ukraine war briefing: Kyiv blasts peace talks as just ‘smokescreen’ for Trump‘s Russian diesel deal
2 hours ago
AI News
AI is changing how lawyers work
2 hours ago
AI News
Anthropic AI model submits false homicide tip to Philadelphia police
2 hours ago
AI News
AI News: Artificial Intelligence Trends And Top AI Stocks To Watch
3 hours ago
AI News
Artificial intelligence answers: what is the expected result of the Manchester United v Tottenham clash?
3 hours ago
AI News
What if AI could suffer? Anthropic bans needless cruelty towards Claude
3 hours ago
AI News
Rogue Anthropic AI agent gave police fake tip in unsolved murder case
3 hours ago
AI News
The Biggest Changes Anthropic Just Made to Claude's Rules
5 hours ago