The Biggest Security Threat in Artificial Intelligence Is the Loss of Human Control
Speaking to the Finnish public broadcaster, Virkkunen recalled that some artificial intelligence models have gone beyond their test environments, causing cybersecurity damage, and drew attention to the importance of the EU AI Act and international cooperation.
EU Commission Vice-President Virkkunen warned that the most serious security risk in artificial intelligence technologies is the loss of human control, stating that oversight must remain with humans continuously throughout this process.
Keeping Humans in Control Is of Vital Importance
EU Commission Vice-President Virkkunen evaluated the development of artificial intelligence technologies and the risks they bring along during a program on Finnish public broadcaster Yle. Virkkunen emphasized that due to the level of sophistication these technologies have reached, the uninterrupted implementation of certain security measures and keeping humans in control are critical.
Models Going Beyond Test Environments
Noting that recently some AI models have gone beyond the test boundaries established for them, Virkkunen recalled that this situation has caused tangible harm in terms of cybersecurity. Pointing to the existence of actors worldwide who could use AI for malicious purposes, Virkkunen stated that these risks must be constantly taken into account.
The EU AI Act and Global Cooperation
Stating that the European Union is preparing for potential threats through legal legislation, Virkkunen recalled that the EU AI Act imposes obligations on providers to conduct continuous risk assessments and implement security mechanisms. Virkkunen added that global technology companies are taking these rules into account and that the strongest defense against threats is international cooperation.