Disclosure

Disclosure in AI governance refers to the requirement for organizations to transparently communicate information about their AI systems, including their capabilities, limitations, training data sources, and potential risks to stakeholders and regulators. For enterprises, robust disclosure practices are critical for building trust, meeting regulatory requirements like the EU AI Act and SEC rules on AI risk, and managing liability exposure. Inadequate disclosure can result in compliance violations, reputational damage, and legal consequences when AI systems cause harm or operate in ways that weren't properly communicated to affected parties.

1 item

ResearchGlobal2026-07-21

32% of Recent arXiv Papers Flag as AI-Written, Exposing Reliability Gaps in Detection Tools Enterprises Rely On

A study by Unsloth scored 12,750 arXiv preprints and found approximately 32% of recent submissions display markers of machine-generated text, up from a pre-ChatGPT baseline of 0.4%. The prevalence varies sharply by discipline, with computer science papers flagging at 65% and mathematics near 0.7%. The study also documents significant limitations in current AI detection methodology, including blind spots and the inability to distinguish lightly edited from wholly generated text.

AI detection academic integrity AI-generated content transparency disclosure content provenance Unsloth