Major Japanese telecommunications firm KDDI signed an agreement Tuesday with Google Cloud Japan in a bid to develop a ¡°responsible¡± AI search service that only shows content that creators have given consent to.

The agreement would allow KDDI to harness Google¡¯s AI assistant Gemini and AI-optimized research tool NotebookLM.

¡°We will promote ¡®responsible AI¡¯ that uses AI ethically, legally, and appropriately, and provide an environment where content providers and customers can use AI safely and securely,¡± the company said in a statement.

KDDI said its AI service, which will be launched in spring 2026, will ¡°protect the rights of content providers.¡±

Ownership and AI is something of a thorny topic ¡ª AI companies, which have trained their products on copyrighted material without creator consent, have faced scrutiny and legal challenges.

In Japan, major newspaper publishers Nikkei, the Yomiuri Shimbun and the Asahi Shimbun filed a lawsuit with the Tokyo District Court in August against U.S. AI startup Perplexity AI, alleging the company had engaged in copyright infringement due to unauthorized use of publishers¡¯ articles, and are pushing for an injunction. Yomiuri claimed that Perplexity had accessed more than 100,000 of its articles between February and June to inform its AI product.

Later in August, Perplexity AI announced plans to implement a revenue-sharing model, giving publishers a portion of its revenue.

The legal challenge is ongoing.

KDDI is hoping to sidestep such issues in its AI offering by seeking permission from third parties in advance. Six news and media firms, including NewsPicks Select, have signed up for the project.

The use of quality sources will help to create a customizable, trustworthy product, KDDI said.

¡°There are growing concerns that the unauthorized use of public content for AI services threatens the sustainable development of journalism and the creative industry,¡± the statement said.

Earlier this month, OpenAI¡¯s shortform video app Sora 2 triggered concern about copyright infringement after the likenesses of iconic Japanese pop culture characters appeared across the platform.

OpenAI subsequently revised the service, seeking permission from copyright holders in advance, after the Japanese government and other stakeholders urged the company to take necessary measures.