AI autonomy on display, but limits unclear

The claim stems from coverage around the Seoul International Aerospace and Defense Exhibition (ADEX 2025), where Korean defense firms showcased AI-equipped unmanned systems described as capable of evolving beyond remote piloting. The exhibition served as the stage for presentations of drone technologies featuring onboard AI intended to assist with mission execution and target engagement.

However, the boundary between AI assisting human operators and AI independently deciding to attack is not established by the available material. Modern defense doctrine in most countries, including discussion around manned-unmanned teaming, keeps a human in the loop for lethal decisions, and no source in the reviewed material states that Korean drones have removed that safeguard.

What supports the claim, and what does not

The part of the claim that holds up is the direction of technological evolution: AI piloting, mission assistance, and intelligent targeting features are being actively developed and marketed for Korean drone platforms, and the exhibition showcased systems described in such terms.

The part that remains unverified is the assertion that AI pilots already write their own operational plans and execute strikes on targets without human authorization. No reviewed evidence confirms autonomous attack authority in operational use, and descriptions at the exhibition did not establish that capability as fielded fact.

ADEX 2025 showcased genuine and substantial progress in AI-driven drone autonomy, but the leap from AI-assisted piloting to fully self-directed operations and attacks has not been demonstrated by the available evidence, making the overall claim Partly True.