Crystal Dynamics Confirms Using Generative AI Tools

According to an interview published by Game Informer and subsequent coverage by Polygon, Crystal Dynamics experience director Jeff Adams described the studio's use of generative AI during early development of Tomb Raider: Legacy of Atlantis. Adams is quoted verbatim by Game Informer: "At Crystal Dynamics, we see AI as a tool that can help our team get to right answers faster." He described using a generative AI tool to "visualize" candidate objects in levels, then moving successful visualizations into the studio's traditional pipeline where the team will concept and build them. Adams added, "we'll make sure that all the finished content in the final game is human-crafted," as reported by Game Informer.

A PR representative intervened during a follow-up question about whether generative assets carried scripts or were purely environmental art, saying - off-camera, per Game Informer - "I think what I'd probably say is I think we've said all we want to say about it now." Observers also pointed to a Steam page disclosure that listed generative AI use for "ideation," a detail highlighted by Polygon and other outlets. Reporting by Kotaku and Gamespot captured the ensuing debate and scrutiny from players and journalists about how and when AI is used and how studios disclose that use.

Industry-pattern observations: studios commonly test generative AI for rapid prototyping and visual ideation rather than final-asset production. Public reporting and interviews in this case align with that pattern: the quoted workflow (use AI to produce a visualization, evaluate it, and then rebuild through the normal asset pipeline) matches documented prototyping use cases across game studios. Several outlets referenced other publishers experimenting with AI-assisted tasks, placing Crystal Dynamics's statements in a broader industry trend rather than an isolated example.

The reaction to the Steam disclosure and the follow-up interview illustrates persistent tensions between developer practices and community expectations around provenance, credit, and creative authorship. Reporting from Kotaku emphasized skepticism about scripted talking points when developers are pressed on follow-ups. For practitioners, this episode underscores the reputational and communication risks that accompany incremental tool adoption: disclosure language, asset-by-asset provenance, and pipeline descriptions matter to players and press.

Observers will likely monitor how publishers phrase Steam or storefront disclosures (for example, whether descriptions move beyond single-word tags like "ideation"), whether studios publish more detailed pipeline notes or credits for assets influenced by AI, and whether third-party tool providers become named in disclosures. Watch for follow-up coverage that documents whether assets attributed as "human-crafted" include underlying AI-origin visuals used during concepting, and whether industry bodies, unions, or platform policies update guidance on AI credits.

Teams adopting generative AI for ideation should expect scrutiny on transparency and asset provenance. Common mitigation approaches include maintaining internal provenance logs, keeping AI outputs clearly separated from deliverable art until human review and reconstruction, and preparing concise disclosure text that communicates the role of AI in the pipeline without implying final-asset generation where none occurred.

Reporting from Game Informer, Polygon, Kotaku, and Gamespot documents that Crystal Dynamics used generative AI as a rapid ideation tool for Tomb Raider: Legacy of Atlantis and that the studio's public comments emphasize human rebuilding for final assets. The story sits at the intersection of practical prototyping workflows and an ongoing public debate about disclosure and creative credit in game development.