Smartphones, which house the most confidential data, should use a security app against potential malware, said research group director for machine learning technology at cybersecurity firm Kaspersky, Vladislav Tushkanov.
“Artificial Intelligence (AI) assistants usually exist in two forms, running on the device and accessed via the internet, and online assistants should not be given your confidential information and their advice should not be trusted in critical situations, as they may provide false information,” said Tushkanov.
Read more: Sindh Braces for Heavy Rains
Accounts should be protected with strong passwords and two-factor authentication, which, if compromised, can leak information given to chatbots, Tuskanov added.
Assistants on the device largely solve privacy issues when they do not require internet access, it was also emphasized.
“However, if these assistants use tools that send queries to third-party services, the privacy threats become similar to the built-in assistants of major smartphone brands.
“Also, due to the limited hardware of smartphones, on-device models tend to underperform and therefore the problem of misinformation and fabricated facts may be more pronounced,” he said.
Read more: Change in jail staff, rats in Bushra Bibi’s cell: Imran Khan details jail conditions
While app providers may use data from chats to improve their models or for other purposes, it was noted that care should be taken not to make conversations too personal, especially when talking to virtual friends or chatbots that offer emotional support.
“The most confidential data we keep on our smartphones can be leaked by a malicious app, AI-powered or not.
“Kaspersky detects threats that appear to be clients of popular AI services, and to protect themselves, users should only install apps they trust, pay attention to the permissions they request, and use a security solution against potential malware,” he concluded.