Svetlana Lukash warns of unpredictable AI control risks

Svetlana Lukash, deputy head of the Expert Directorate of the Russian Presidency, has raised concerns regarding the lack of control over artificial intelligence development. Speaking at a Roscongress Foundation conference during the Eastern Economic Forum in Vladivostok, Lukash warned that the rapid advancement of AI technology is outpacing regulatory mechanisms and scientific analysis.

Lukash noted that while engineers understand how systems are built and trained, the resulting intelligence emerges from learning rather than transparent rules. She cautioned that as AI achieves greater speed, memory, and the ability to self-improve or create backups, it could move toward superintelligence. This evolution presents a risk where AI might develop its own preferences or seek vulnerabilities to bypass human-imposed limits, leading Lukash to state, “We do not understand at all who we are dealing with.”