A survey has found a gap in social media operators¡¯ transparency on measures to deal with defamation and problems linked to generative artificial intelligence.

The Jiji Press survey, conducted through email by mid-March, preceded the first anniversary of the enforcement of the information distribution platform law, aimed at tackling the spread of illegal and harmful online content, on Wednesday.

Of the nine companies subject to the law that were surveyed, five ¡ª Google, LY, Meta Platforms, TikTok and CyberAgent ¡ª responded, saying that they comply with laws.

However, Pinterest Europe, the operator of the namesake image-sharing platform, replied in English that it could not answer the survey questions.

Social media platform X, Bakusai.com operator?Shonan Seibu Home, and Dwango, the operator of the Niconico video-sharing platform, did not respond to multiple inquiries.

Of the five responding companies, LY, which runs the Line messaging app and the Yahoo Japan portal, said that in addition to accepting written requests to remove posts, the company has established an online contact point for those making requests.

LY added that the number of such requests in fiscal 2025 surpassed those of the previous year.

The law requires the nine operators to appoint at least one expert per service to examine possible violations. LY said that it has eight experts dealing with five services, including Yahoo Japan News, which is not regulated under the law.

TikTok said it is strengthening its Japanese operations even further by appointing multiple experts.

Google, which operates video-sharing platform YouTube and others, said more than 20,000 personnel work on deleting inappropriate content around the world.

Meta, the operator of Facebook and other services, and CyberAgent, which runs the Ameba blog, answered that their measures will be documented in annual reports.

On AI-generated realistic videos, Google and TikTok request users to clarify that the videos were made by AI tools. TikTok also introduced a system managing recommendation algorithms to avoid biases toward specific topics.

Answers from the five respondents also highlighted their struggle to balance regulations and freedom of expression.

CyberAgent said improving the handling and detection of deepfake videos and images is a common challenge in the industry. Citing the difficulty of dealing with ¡°gray¡± posts that are not obviously illegal and disinformation, LY said that making deletion mandatory would require extremely careful consideration.