Sega will use AI in game development - but only where it makes clear operational sense
Sega says it will adopt AI in development, but selectively. In a Q&A following its Q2 results, executives said the company will prioritize efficiency improvements "such as leveraging AI," while acknowledging "strong resistance in creative areas such as character creation."
The stance is simple: AI where it removes grind and speeds up workflows, human ownership where it defines the game. Sega says it will "carefully assess appropriate use cases" and focus on streamlining processes rather than chasing ever-larger productions.
Why this matters for IT and development teams
- Signals a pragmatic, ops-first approach to AI: think build acceleration, tooling, localization, and QA support-not final character art or narrative.
- Sets expectations for guardrails in creative pipelines. You'll need policy, audits, and human review baked into your toolchain.
- Supports hybrid staffing: engineers and artists who can work with AI tools without outsourcing authorship to them.
Hiring and policy are shifting on the art side
A Japanese report this week noted some studios now ask art candidates to draw during interviews to verify they can actually produce work, not just pass off AI outputs. Teams reportedly hired applicants who relied on generative tools and found productivity issues later.
There's also pressure from some leaders to ask whether "AI is good enough" for certain roles, or if they should prioritize candidates skilled at prompting. Expect more explicit portfolio checks, live skills tests, and disclosure requirements in hiring flows.
A middle ground is taking shape
Arrowhead Game Studios' CEO Shams Jorjani recently argued for nuance: avoid AI that replaces creativity, adopt AI that kills admin and busywork. His example was using AI for receipts and transcription, not to write the game.
That mirrors Sega's framing. Efficiency gains are fair game; authorship stays human.
Experiments show promise-and friction
Ubisoft recently demoed "Teammates," where players give natural-language commands to AI squadmates. Early impressions suggest it can work, but there are hurdles before players accept it and before a full release makes sense.
Translation for your roadmap: prototypes are convincing, but reliability, UX clarity, and trust need solved before shipping at scale. A measured rollout with evaluation gates is smarter than a blanket rollout.
Practical playbook for "appropriate use cases"
- Target low-risk, high-grind tasks first: build orchestration, content tagging, asset deduplication, localization drafts, speech-to-text, bug triage, log analysis, test case generation, and tooling docs.
- Keep authorship human for creative pillars: character art, worldbuilding, narrative beats, voice performance, signature animation. If it defines the IP, keep humans in the driver's seat.
- Set policy and disclosures: require staff to disclose AI-assisted work; watermark intermediate assets; track provenance in your asset DB.
- Govern data and licensing: restrict training and prompts to approved datasets; block unvetted external endpoints; ensure you have rights for anything that touches production.
- Security and privacy by default: no secrets in prompts; use VPC-hosted or enterprise endpoints; add guardrails and red-teaming for prompt injection and data leakage.
- Quality gates: define success metrics and eval sets per use case; require human review for anything visible to players; keep rollback paths to manual workflows.
- Hiring and upskilling: use live skills tests for art and code, add AI-use declarations, and train teams on safe and effective tool use.
- Compliance and risk: map use cases to an AI risk framework and run lightweight reviews before scaling.
For a solid structure to assess risk and controls, see the NIST AI Risk Management Framework here.
Bottom line
Sega's position reflects where many studios are landing: use AI to speed up production, not to define the art. That middle path delivers tangible gains without compromising creative identity.
If you're formalizing this in your studio, start with internal tools and production support, write clear "AI-eligible" rules, and train your team to use the stack well. Need structured upskilling? Browse role-based options here.
Your membership also unlocks:
