Legal Limitations of Algorithmic Analytical Tools Against Disinformation


Publikace nespadá pod Fakultu sociálních studií, ale pod Právnickou fakultu. Oficiální stránka publikace je na webu

Název česky Právní limity algoritmických analytických nástrojů proti dizinformacím

KASL František

Rok publikování 2018
Druh Další prezentace na konferencích
Fakulta / Pracoviště MU

Právnická fakulta

Popis The spread of fake news, disinformation and anti-democracy propaganda through the online media and social networks is an acute problem. Particularly challenging is the need for credible, reliable and scalable approach to swift identification and assessment of illegal content that is necessary for legal process of takedown or removal by the responsible internet service provider and also for subsequent criminal prosecution of the offender. The author of the contribution is a member of an interdisciplinary research team of IT experts, security analysts and lawyers focusing on developing a method of recognition, detection and analysis of manipulative techniques on pro-Russian online media outlets through automated analytical software based on machine learning and big data mining [Project MUNI/G/0872/2016 Manipulative techniques of propaganda in the age of internet]. The use of an analytical tool for recognition of propaganda disinformation, as the one that is being developed in the aforementioned project, evokes several important legal questions that need to be considered in order to properly assess the potential viability of this approach in the struggle against propaganda manipulation. The particular issue at hand that is to be discussed in this contribution are the parameters of the legal framework for this analytical software and the ensuing legal limitations of this tool against disinformation. This form of data analytics should provide evidence for criminal or civil liability procedures, but such application requires critical probes into objectivity and impartiality of the algorithmic tool. To pursue this issue to its core, it is suitable to follow to the basis of Wiener´s cybernetics and identify the limits of compatibility between application of law and algorithmic expression of abstract terms. Patterns, correlations and axioms need to be identified and verified in order to imitate legal or other abstract expert assessment. Such inputs into the algorithmic decision making need, however, to be subject to adequate scrutiny and verification, in order to prevent from flawed or biased foundation for interpretation and application of law. This challenge can also be found in the background of many legal frameworks aimed at establishment of tools necessary for appropriate regulation and moderation of online conduct and data flows. The contribution shall discuss the presented challenges in greater detail as well as provide the summary of findings obtained on the current pursuit towards mitigating the limitations emerging in their consequence. The aim is to provide a framework of an appropriate setting of the analytical algorithm that would satisfy the legal prerequisites for producing supporting evidence in a process of takedown or removal by the responsible internet service provider or in a criminal prosecution of the offender.
Související projekty:

Používáte starou verzi internetového prohlížeče. Doporučujeme aktualizovat Váš prohlížeč na nejnovější verzi.