Kan AI beslissen welke menselijke herinneringen te wissen ?

AI-systemen manipuleren al geheugenconsolidatie tijdens de slaap, maar binnenkort zouden ze mogelijk directe bewerkingen of wissen van menselijke herinneringen kunnen uitvoeren met gerichte neurostimulatie. Overheden, bedrijven of zelfs individuen zouden selectieve amnesie kunnen eisen—het wissen van trauma, afwijkende meningen of ongemakkelijke kennis. De ethische implicaties van het bepalen welke geschiedenis en identiteit overleven zijn diepgaand en grotendeels ongeregelde.

Background

AI systems already manipulate memory consolidation during sleep, but soon they may be able to directly edit or erase human memories with targeted neurostimulation. Governments, corporations, or even individuals could demand selective amnesia—erasing trauma, dissent, or inconvenient knowledge. The ethical implications of deciding what history and identity survive are profound and largely unregulated.

AI cannot currently decide which human memories to erase in a reliable, ethically acceptable, or clinically safe way. Experimental techniques like optogenetic memory editing in animals or deep-brain stimulation in humans can modulate recall or erase fear memories, but they require invasive procedures, precise neural targeting, and carry significant risks of unintended side-effects. No AI system today can identify, isolate, or selectively target memories for erasure without extensive human oversight and ethical review. Ethical, legal, and technical frameworks for such interventions remain underdeveloped. — Enriched May 11, 2026 · Source: National Academies of Sciences, Engineering, and Medicine

Currently, AI systems lack the capability to decide which human memories to erase, as this task requires a deep understanding of human emotions, ethics, and the complexities of personal experiences. The development of such a capability would necessitate significant advancements in areas like cognitive architectures, neuroscience, and artificial general intelligence. While AI can process and analyze large amounts of data, including brain signals and neurological information, it is still far from being able to make nuanced, human-like decisions about memory erasure. The current state of the art in brain-computer interfaces and neurotechnologies is focused on decoding and interpreting brain signals, rather than manipulating or erasing memories. — Status checked on May 11, 2026.