AUTHORSHIP IN WORKS ASSISTED BY ARTIFICIAL INTELLIGENCE: CONTEMPORARY CHALLENGES AND THE ADVENT OF INTELLIGENT AGENTS

Abstract

This article analyzes contemporary challenges related to the authorship of academic and professional works assisted by artificial intelligence (AI), distinguishing conversational chats from intelligent agents capable of planning and executing complex tasks. A bibliographic and documentary review was conducted of publications, institutional guidelines, normative acts, and documents addressing academic ethics, algorithmic governance, and content licensing, with emphasis on Brazil. The results indicate that human-AI collaboration may take assisted, autonomous, or agent-driven forms, but operational autonomy does not eliminate human responsibility for the outcome. Manus AI is examined as a reference case of a work-oriented tool that combines dialogue, planning, browser and file-system use, and delivery of artifacts, a combination still uncommon among general-purpose tools. The study concludes that integrity depends on human critical review, process documentation, source verification, and proportionate disclosure of technological contributions. The analysis further argues that this architectural distinction should guide editorial policies, transparency protocols, and authorship criteria for research assisted by AI.