Three months ago, Habr introduced a rule prohibiting the publication of content generated or edited by artificial intelligence. Despite this policy, automated verification tools are absent, leading to the continued appearance of AI-generated posts on the platform.
Analysis using the GPTZero detector revealed that multiple recent articles from corporate blogs, including AvitoTech and Timeweb Cloud, show high percentages of AI-generated content. Conversely, RUVDS posts tested showed no signs of AI usage.
The author notes that moderation remains inconsistent. Specific articles flagged by users as AI-generated, such as those regarding PostgreSQL and EvertyDesk, remain on the platform despite user complaints and reports submitted to the moderation team.
The platform currently utilizes free tools like GigaCheck and Yandex Neurodetector for content verification, which the author claims are ineffective. Implementing more accurate, paid detection services would require significant monthly investment, which the platform has not committed to.
The author highlights a case of a user who published 12 AI-generated articles within 20 days after passing the initial moderation process via the platform's sandbox. There is currently no automated system to filter or label such content for users.
