ICML Implements New Policy on Generative AI Use in Research Submissions, Mandating Disclosure and Author Responsibility
The International Conference on Machine Learning (ICML) has introduced a new policy for 2027 regarding the use of generative AI tools in research submissions. This policy permits the use of AI tools, such as large language models, in preparing papers, provided that their use is explicitly disclosed. Authors are now required to include an 'AI Disclosure Statement' at the end of their paper, detailing how generative AI tools were utilized for research or writing tasks, how AI-assisted outputs were evaluated, and any material use of AI relevant to understanding or reproducing the work. The policy emphasizes that authors retain full responsibility for all content, including the correctness and originality of scientific claims, experimental results, code, and citations, regardless of AI involvement. Routine editorial assistance, like grammar correction or formatting, does not require disclosure unless it materially alters the intellectual or scientific content.
