argument: Notizie/News - Digital Governance
Source: To Vima
Katerina Mavridou argues that artificial intelligence requires effective accountability when corporate systems influence decisions affecting citizens. Her opinion opens with recommendations released by the Australian government on 10 November 2026 following an incident involving an OpenAI AI agent accessing government websites. According to the account, the government considered the information received inadequate and announced plans for independent checks and mandatory incident reporting. The text expressly states that a leak of individual medical records has not been established. Mavridou argues that companies can exercise substantial social influence without offering affected people corresponding opportunities for scrutiny. She questions how responsibility should be allocated among developers, organisations deploying AI and people accepting automated decisions.
Mavridou uses the 2023 SCHUFA case to illustrate concerns about accountability when automated credit scores shape lending decisions. She refers to the GDPR and the AI Act as European safeguards, while questioning whether affected citizens can effectively exercise their rights. In her analysis, public authorities’ dependence on privately supplied technology can weaken their practical ability to scrutinise decision-making. She also recognises AI’s benefits and argues that excessively rigid regulation can advantage established businesses over smaller innovators. Her proposals include knowledgeable supervisors with time and authority to disagree, public services capable of examining AI use, and accessible review procedures. Mavridou ultimately calls for public debate about citizens’ ability to understand, scrutinise and challenge consequential AI decisions.