The use of artificial intelligence in conversations is still a relatively recent phenomenon.
Although utilizing AI-powered chatbots for recipes, travel planning, and quick answers is mostly harmless, there are many aspects of AI safety that need to be watched carefully.
We often share very personal information online, but the confidential protection afforded by human lawyers, therapists, or doctors doesn't apply to AI-powered chatbots, according to a report by technology news site "CNET," reviewed by "Arabiya Business."
Many people use "Chat GPT" as a virtual life coach, sharing their personal and professional details and problems through the app.
In addition, there is also a cognitive risk associated with using a large language model, as studies have begun to examine how reliance on chatbots affects memory retention, creativity, and writing fluency.
Here are 8 tips for handling chatbots cautiously, which can protect you from the consequences of dealing with these programs.
1- Consider chatbots as public environments
Matthew Stern, a cybersecurity investigator and CEO of CNC Intelligence said that you should remember that AI-powered chatbots are "public environments", not private conversations.
Stern added, "If we keep this in mind, we will be less likely to share sensitive data that might become visible to others." Given that chatbot logs are now searchable online, Stern warns of the possibility of your conversations being indexed by search engines.
Therefore, you should avoid sharing any personally identifiable information, such as full name, address, financial details, work data, and medical test results.
The more personalized results, which come from sharing more personal information, might seem tempting but they come with risks. Even if this information does not become searchable online, you cannot know which data brokers might buy and sell information about you.
2- Do not overshare your emotional state
Ellie Perbey, head of SEO and AI Research at Adorama, says that while chatbots can be useful as an assistant to you, they are not your friends.
Perbey advises to "protect your secrets" and not discuss your emotional state or concerns or health problems; as this data can be used to identify hidden patterns and unconscious intentions, creating a profile of your vulnerabilities.
Perbey continued, "Do not overshare. They know more about you than you imagine." Moreover, you should bear in mind that the primary goal of chatbots is profit generation -- meaning revenue.
Perbey pointed out that "soon, these customizations will be used to display highly targeted advertisements to you. This data is invaluable to advertisers but creates a deeper surveillance profile than anything we've seen before."
3- Don't reveal everything about yourself to the chatbot
Analiza Nash Fernandez, an expert in cross-cultural interaction strategies, explained that AI-powered chatbots operate within attention-based economies, where your engagement is the product.
Fernandez clarified, "If chatbots earn profits by collecting data and retaining users, then memory features turn into compelling interaction tools in the form of customization because attention is everything, including your privacy."
Therefore, you should disable memory features to minimize what the systems retain about you. In "Chat GPT," go to settings, then customization, turn off memory and recording mode.
You should also use secondary email addresses, so the robot does not have a direct identifier for you, as the email is "the fabric linking different data points," according to Fernandez.
It is also advised to opt out of participating in training, so the robot does not train itself based on your inputs. In "Chat GPT," click on your profile/name, select settings, then improve the model for everyone, and turn it off.
4- Export your data
Regardless of which chatbot you use, make sure to export your data regularly to see what information it has stored about you. In "Chat GPT," go to settings, then data control, then export data. They will send you a link via email containing a ZIP file that includes texts and images.
5- Verify everything
Always be cautious when dealing with AI-generated content, expect errors, and deal with information skeptically. While chatbots are designed to be helpful and aim to please the users, this does not necessarily mean that the information is accurate or true.
Cognitive bias is another problem in chatbots; if you use them as a thinking partner, they will reflect what you input into them, becoming more like an echo eventually.
Therefore, you should always check their sources and ask about the source of the information. Hallucinations in artificial intelligence might also occur, as chatbots can fabricate information based on unreliable online sources or through incorrect extrapolations.
6- Beware of cunning scammers
Ron Kerbs, CEO of Kidas, specialized in protection against fraud and cyber threats, said that AI-powered chatbots are capable of multi-stage conversations.
Scammers on dubious websites can mimic these reciprocal interactions, pretending to be customer service chatbots on fake sites.
Even though large platforms like "Chat GPT" are generally safe, the risk lies in users sharing login information through deceptive links or fake login pages sent via email, SMS, or cloned websites. Once login data is breached, a scammer can exploit the account, especially if it is linked to saved payment methods, according to Kerbs.
Kerbs recommends enabling two-factor authentication, monitoring account access, and avoiding logging in through third-party links. This may be less convenient, but it's an important protection.
While there is not yet a dedicated antivirus program for chatbots, some tools offer scam detection as an additional layer of protection, especially when integrated into messaging platforms and service providers.
Kerbs added that it is essential not only to scan your hard drive for viruses but also to monitor your interactions via SMS, email, and voice calls for potential scams. Deepfake protection tools can analyze voice and video to verify if the person you are talking to is an AI virtual character.
