Identifying signs of comments created by AI to incite and attract public opinion

Suspicion when many accounts repeat the same message

Under many articles with controversial content on social networks, it is not difficult to encounter a series of comments appearing in just a short time with similar expressions, repeating the same message.

At first glance, this can be seen as a community reaction. However, in many cases, behind those "waves" of comments may be content made by artificial intelligence (AI). This increases the risk of spreading false information, slander and incitement in cyberspace.

Talking to a reporter from Lao Dong Newspaper, network security expert Ngo Minh Hieu (aka Hieu PC) - Project Director for Anti-Fraudulent Fraud, member of the Vietnam National Cyber Security Association - said: "There is no single sign enough to confirm that the comment was created by AI. However, it is possible to suspect when many accounts post almost simultaneously, repeating the same message or sentence structure, using smooth but general language, responding at unusual speeds and together focusing on leading public opinion in one direction.

Experts note that readers should also click on the comment account to check. Virtual accounts are usually newly created; personal information is rudimentary or inconsistent; avatars show signs of copying; very few everyday articles; many images are posted at the same time; low interaction or mainly from other suspicious accounts.

In addition, expert Ngo Minh Hieu also stated the fact that comments are often very short, may be edited by humans, mixed between real content and AI content or operated by both real people and automated systems.

AI detection tools still have the risk of making wrong conclusions, while data on IP addresses, devices and login history is only held by the platform. Therefore, verification and processing must be based on the overall false content, coordination behavior, account network and consequences caused, and cannot just rely on the conclusion'created by AI'. The detector should save the link, screenshot, time and account information to send to the platform or functional agency" - expert Ngo Minh Hieu recommended.

AI makes forgery and spread of information more sophisticated

Expert Ngo Minh Hieu said that in the process of supporting people and businesses, he has received many cases of personal data leakage leading to images and information being used illegally on social networks.

According to Mr. Hieu, some businesses are harassed by bad actors using AI to impersonate images of leaders, then create defamatory content, distort or cut and paste with untrue information. Some other cases are also exploited for illegal advertising, extortion or committing fraudulent acts in cyberspace.

Protecting information, personal data and business data is becoming particularly important in the context of increasingly developed AI technologies," Mr. Hieu emphasized.

This expert also believes that currently AI technology can create fake images, videos and voices with a very high level of authenticity. Therefore, users need to be more careful with content appearing on social networks, especially information related to celebrities, businesses or cases that are attracting public attention.

Users should verify information through official channels of related individuals, organizations or press agencies, functional agencies before sharing" - Mr. Hieu recommended.

Making a record to preserve evidence when being defamed on social networks

In the context of many cases of slander, defaming and spreading false information on social networks being "heated up" by a series of provocative comments, Lawyer Tran Xuan Tien - Head of Dong Doi Law Office (Hanoi Bar Association) - said that to protect rights, victims need to implement a methodical and scientific legal process.

First of all, as soon as defamatory information is detected, it is necessary to immediately collect and preserve electronic evidence by taking photos, recording video screens to store articles, insulting comments and URL links of violating accounts.

The next decisive step is to coordinate with the Bailiff Office to make a written record recording these evidences on the online environment. After having the written record, the victim proceeds to send a petition for administrative handling with evidence to the functional agency.

Finally, depending on the extent of damage and the seriousness of the act, the victim can file a civil lawsuit in court to request compensation, or send a denunciation of crime to the Investigation Police Agency to request criminal prosecution in accordance with current procedural law.

Read the original here