The growing reliance on AI chatbots is raising a major privacy. New research from DuckDuckGo shows that many users treat AI systems as a safe and personal space for conversation, revealing information they probably wouldn't share even with people in their immediate environment.
Personal conversations are increasing
According to the survey, nearly one in three users has told a chatbot something they wouldn’t reveal to friends, family, colleagues or even healthcare professionals. The percentage is even higher among users who say they are particularly ardent supporters of AI, reaching 56%.
See also: DuckDuckGo: Introduces its own AI Chat that focuses on security
The difference compared to a traditional search engine is significant. A search usually consists of a few words, while a conversation with a chatbot can develop into a detailed description of a problem, a personal situation, or even feelings and thoughts.
This way, AI can gain a much more complete picture of the user: not only what they are looking for, but also why they are looking for it, how they think, and what is the context behind their questions.
Large information gap for data
The survey, which surveyed nearly 2,000 adults in the U.S., also reveals significant knowledge gaps. 53% of respondents said they were unaware or unsure that many chatbots can use conversations to models their, depending on the service and its settings.
Even fewer were aware that conversations with AI could, under certain legal conditions, be requested by authorities through judicial proceedings. Just 25% said they were aware of this possibility.
When participants were informed about using conversations for AI training, 58% said they felt uncomfortable, while 30% reported feeling “very uncomfortable.”
See also: Microsoft outage affects Bing, Copilot, DuckDuckGo, and ChatGPT
Parents show more trust
The difference between parents and users without children at home is also interesting . Among parents who use AI, 43% said they have shared something with a chatbot that they had never disclosed to another person.
For users without children, the corresponding percentage is limited to 25%. This difference suggests that the perception of chatbots as available “interlocutors” may significantly influence how personal the questions are.
The trend is reinforced by the way AI services are marketed. The more a chatbot is presented as an assistant, advisor, or digital conversationalist, the easier it is for the user to forget that they are interacting with a technological service and not with a person bound by a human trusting relationship.
Limited trust in large companies
Despite increased use, trust remains low. Respondents are slightly more positive about large AI companies than they are about the US government when it comes to protecting their data.
However, 39% said they did not trust large AI companies, while the corresponding figure for the government reached 48%.
🔒 Protect your privacy with Proton VPN
Swiss VPN from the creators of Proton Mail — strict no-logs policy, strong encryption, and built-in NetShield that blocks ads, trackers, & malware.
- ✔ No-logs, based in Switzerland (except 14-Eyes)
- ✔ NetShield: blocks ads, trackers & malicious domains
- ✔ Covers all devices — free version available
The link is an affiliate link — SecNews may receive a commission at no additional cost to you. It does not affect the independence of our article writing.
See also: DuckDuckGo browser blocks ads on YouTube
Privacy starts with the next question
The most important takeaway is that privacy is not necessarily a lost battle. Users can change their behavior now by limiting the information they input into chatbots and carefully reviewing each service's policies.
In practice, it is advisable to avoid sending passwords, financial details, sensitive documents and information that can directly lead to the identification of a person. At the same time, it is worth checking the settings of history and data usage and preferring services with clear and transparent practices.
In the age of AI, privacy doesn't just depend on what we've already shared online. It also depends on the next question we type.
