Activities:

A tool based on artificial intelligence assists RRT in ensuring clean internet

2022 April 11 d.

Since the beginning of 2022 Communications Regulatory Authority (RRT) started using a tool developed by company “Oxylabs” based on artificial intelligence for the search for harmful content on internet. More than 288 thousand of Lithuanian websites were checked using this tool. The potentially illegal and harmful or harmful content identified on these websites was sent for further assessment of RRT specialists. After RRT specialists investigated every notification, it was established that 19 websites possibly breached Lithuanian national or European Union laws: 8 reports were forwarded for further investigation to Police Department, 11 – to the Office of the Inspector of Journalist Ethics and 2 pre-trial investigations were initiated regarding the reports. For comparison: 80 reports received from internet users were forwarded for further investigation to Police Department and 31 report– to the Office of the Inspector of Journalist Ethics during the entire 2021.

“Up to now the success of fighting against harmful and illegal content on internet depended upon the benevolence of users, since they used to send us reports related to pornography, child sexual abuse material , violence and etc. using our hotline “Švarus internetas” (“Clean Internet”). In order to make this procedure more efficient, at the end of 2020 during “GovTech Lab” idea challenge session we have created a challenge, the purpose of which was to create a tool, which would find child sexual abuse material or pornography on Lithuanian websites. The challenge was won by company “Oxylabs” which suggest the best solution,” stated Vaidotas Ramonas, the Director of Digital Service Department of RRT. “Even though the voluntary reports, received via our hotline, is a method used all over the world that is also very useful, the intention was not to be dependent solely on these reports. We hope that the artificial intelligence technology, applied simultaneously with the ordinary image hashing, will allow us easier to identify the content, both that was already identified before and that was recently uploaded.”

According to V. Ramonas, the goal was to create a tool for recognition of information related to sexual abuse material or pornography, since reports of this kind are the most frequent every year. For example, almost half of the reports received during 2021, concerning which further actions were undertaken, were reports about child sexual abuse material.

Artificial intelligence was invoked

The winner of the tender “Oxylabs” started the creation of automated pro bono tool for recognition of illegal and harmful content on internet. The works lasted several weeks, later the developed tool was tested, artificial intelligence was invoked for the trainings, defects identified during the testing were eliminated and improvements were supplemented. Since 2022 RRT started using this tool in the everyday operations of the institution.

“For the implementation of this task we have invoked data collection technology and information recognition methods based on artificial intelligence, because of which the tool we have create is able to scan thousands of different pages every day and recognise the potentially illegal and harmful images. Upon finding something the tool forwards it to RRT hotline for further assessment by the specialists – just like internet users,” such explanation was provided by Juras Juršėnas, the Executive Director of “Oxylabs”.

According to J. Juršėnas, the company has a pro bono partnership programme, the purpose of which is to invoke the big data technologies developed by “Oxylabs” for social and public purposes. One of such examples: at the start of the pandemic the company provided the necessary tools to scientists from Stanford and other leading universities, engaged in the research of the pandemic.

 In order to prevent dissemination of harmful and illegal information RRT will continue to use both the notification platform of hotline “Švarus internetas” (“Clean Internet”) and recognition tool created by “Oxylabs”. Since they complement each other, they will enable to identify illegal and harmful content on Lithuanian internet websites as well as to undertake actions in order to ensure that such content is eliminated as soon as possible. Moreover, RRT intends to share the good artificial intelligence application experience with similar institutions in other countries.

The nature of reported information is changing

During 2021 RRT received a record number of reports about illegal or is harmful to minors via internet hotline “Švarus internetas” (“Clean Internet”) – 3558 reports. After the investigation every report, further actions were undertaken regarding 380 cases.

As the military invasion of Russian Federation into Ukraine started, the nature of reports changed. Before it most of the reports were regarding distribution of child sexual abuse material, whereas now most of the reports concern disinformation, war propaganda, inciting discord or disseminating false information.

RRT reminds that internet users are encouraged to report illegal (prohibited to post) or harmful (having negative impact upon minors) content found on internet, the URL which is https://www.svarusinternetas.lt/. The reports can also be submitted anonymously. The instructions on how to report about noticed illegal or harmful content on internet as well as what is the further course of investigation of such reports can be found on this video: https://www.youtube.com/watch?v=TXRBMU4sfOk. For more information about safe and responsible  internet behaviour, please visit the websites www.draugiskasinternetas.lt and www.esaugumas.lt.

Updated on 2022-04-13