85.8% of Japanese game developers use generative AI, a notable rise in one year

Japan's CESA has presented preliminary data from its new industry report, the CESA Video Game Industry Report 2026. According to the data, 85.8% of game developers in Japan use generative AI at least occasionally, and 63% use it regularly. Only a relatively small share of respondents do without these tools entirely. The preliminary version of the report was distributed at Tokyo Game Show 2026. CESA organizes the show and brings together companies from the Japanese game industry, including Capcom, Sega and Konami.

Beyond the share itself, the survey shows how studios organize their work with AI. The management measure companies named most often was having people review, correct or oversee work done with the help of AI. Other approaches include a limited set of approved tools and rules that prohibit using generated output without modification.

A notable rise in one year

In last year's CESA report, about 51% of the Japanese game companies surveyed said they used AI or generative AI in development. Now 85.8% of developers work with generative models at least occasionally. The figures are hard to compare directly: last time the survey was about companies, this time it is about developers. Even so, the increase looks substantial. CESA plans to publish the full report, with a description of the respondents and methodology, in early December; until then, these are preliminary figures.

CESA member companies named improved workflow efficiency and higher productivity as the main expected benefits of AI. Slightly less often, they mentioned shorter development times and lower costs for developing and maintaining games. Judging by these answers, the industry primarily expects generative models to save resources.

What counts as “using AI”

In statistics like these, “using generative AI” does not necessarily mean that generated characters, dialogue or voices end up in finished games. The category can cover much of a studio's internal work:

• programming assistance;
• ideation and documentation;
• localization and testing;
• research and prototyping;
• asset processing.

In CESA's 2025 report, image and visual content generation was mentioned most often. It was followed by story and text generation, and then by programming assistance. At the time, about 32% of companies said they used AI in developing their own game engines. For artists and technical specialists, this means that studios use AI both as an image generator and as part of their production pipeline tools.

A similar picture outside Japan

Outside Japan, the results are similar. A study commissioned by Google Cloud surveyed 615 developers from the US, South Korea, Norway, Finland and Sweden; 87% of them said they use AI agents in their work. Participants found AI most useful in testing, localization, code assistance and multimodal data processing. At the same time, respondents also named unresolved issues: intellectual property, data rights, implementation costs, training data, and the difficulty of assessing whether investments in AI pay off.

Judging by CESA's data, Japanese companies are responding to similar risks by making human oversight a central part of their policies. CESA Managing Director Tsutomu Masuda noted that the survey shows broad and steady adoption of generative AI in development. At the same time, he stressed the need to address copyright and other rights issues while balancing technological experimentation with the protection of creators.

Major technology and entertainment companies are articulating a similar position. Microsoft has published a draft code describing the expected behavior and values of its AI models. The document emphasizes human oversight, dignity and autonomy, and sets out the principle that access to information does not automatically grant the right to reproduce it. For now, it is a draft rather than an active standard. Sony says AI should enhance the work of PlayStation Studios while keeping people at the center of the process. The company currently uses it in QA, software development, animation, motion capture data processing and routine production tasks, with no emphasis on replacing artists and actors with generated content.

What this means for the industry

CESA's data fits into a broader trend: generative AI appears to be becoming an ordinary production tool. The Japanese industry is often considered conservative when it comes to new production technologies. If the preliminary figures are confirmed in the full report, it appears to be on par with studios abroad in terms of AI adoption. This is likely linked to pressure on budgets and schedules: studios name saving time and money among their main expectations of AI.

For professionals, this may change job requirements. If AI output cannot be used without review, skills in reviewing, refining and quality control become more valuable. Studios will likely introduce more internal guidelines, lists of approved tools and data provenance requirements. Companies that quickly put such rules in place will probably gain an advantage; things may be harder for those that use AI without a clear policy, especially if legal questions start being settled in court.

The survey does not settle the debates over art, law and employment. More details, including the makeup of the respondents, will become known after CESA releases the full report in early December.

What is driving this rise of AI in Japanese studios: staff shortages, budget pressure, the availability of ready-made tools?