Martine Saunders is business development and client relations director at the video game data and research firm Fancensus.
This article is part of AI Week.
AI is fast becoming one of the most influential intermediaries between video game publishers and their audiences, with 55% of Fancensus’s gamer panel admitting to having used AI to source game-related information. The Game Awards 2025 offered an interesting, specific opportunity for a case study to demonstrate how AI platforms are starting to shape discovery, hype, and messaging.
A new bridge between brands and players
Fancensus’s case study (AI Engine Observations – The Game Awards ’25, first published December 19, 2025) tracks how three leading AI chat platforms (ChatGPT, Gemini, and Perplexity) responded to questions about The Game Awards 2025 before, during, and after the live broadcast, revealing that AIs are no longer a curiosity, but a key gateway through which gamers learn what’s been announced, who won, and which titles are creating a buzz. By querying each platform with a consistent set of questions over several days, the study shows that AI does more than just repeat headlines, it filters, prioritises, and frames the event and the brands that featured.
Methodology: treating AIs like media channels
The case study treated AIs much like another media source, running over 2,000 structured questions across three distinct periods: pre-show rumour tracking, live-show reaction and announcements, and post-show consolidation. A single rumour-focused question was asked regularly in the run up to the show, with nine further questions covering announcements, winners, platform-specific line ups, and audience sentiment cycled every five minutes during the livestream itself, continuing regularly but less frequently after the event. Fancensus’s final report goes into more granular details, but the high-level trends and directional findings are enough to understand the overarching implications.
Before the show: rumours, leaks, and missed calls
In the days leading up to The Game Awards 2025, all three AI platforms surfaced a broad spread of rumoured titles, consistently latching onto highly discussed topics such as Resident Evil Requiem, as well as other notable high-profile franchises. Yet the hit rate on genuine reveals remained modest, and each AI frequently blurred the line between “announced,” “nominated,” and “wishlisted,” underscoring how speculative conversation can easily bleed into perceived fact once processed through an AI interface. For publishers in particular, this pre-show period highlights both opportunity and risk, as legitimate teasers can travel quickly through AI-mediated chatter, but so can inaccurate expectations and details.
During the show: who the machines decided were winners
Once the livestream began, the three AI platforms quickly converged on a handful of defining titles, with Clair Obscur: Expedition 33 unsurprisingly emerging as the dominant reference point across all questions and platforms. Frequent queries about “most exciting,” “biggest surprise,” and “what people liked” repeatedly elevated the same cluster of games, meaning the AIs effectively reinforced a secondary, AI-driven lineup, layered on top of the actual, official running order. The timing of queries and where within the livestream a title featured had some influence, as titles that featured earlier in the show had more time and therefore more opportunity to be captured and resurfaced, strengthening their presence in subsequent AI responses.
After the show: consolidating the messaging
In the 48 hours following the end of the show, AI responses shifted from real-time capture to narrative and messaging consolidation, with the same platforms now positioning certain games as emblematic of the show’s legacy. Surprise reveals and the polarising Highguard finale were amplified, while some award winners received less AI attention than their trophy haul would suggest, indicating that memorability in AI outputs is driven as much by discourse intensity and distinctiveness as by formal recognition. Mentions of personalities followed a similar pattern, with hosts, headline performers and standout winners repeatedly utilised as shorthand for the show’s cultural impact.
Different AIs, different editorial instincts
Despite drawing on overlapping pools of press, social, and video sources, the three platforms exhibited distinct “editorial” behaviours. One relied heavily on a small set of reference sites, particularly Wikipedia, while another distributed citations across a far broader mix of specialist outlets, general news, technology media, and user-generated platforms such as Reddit and YouTube. Additionally, although fragmented by game/publisher, there were a few official sites referenced, therefore accuracy and completeness of details on owned sites are an important consideration for publishers wanting to ensure their specific messaging cuts through. These differences led to subtly different hierarchies of games and publishers listed across the platforms, meaning a title’s prominence can vary significantly depending on which AI a player happens to ask. It’s also worth noting here that smaller, less obvious sources can have a strong representation; identifying those sources that punch above their weight could help to push messaging and increase visibility.
Strengths, blind spots, and occasional hallucinations
The case study indicates that, at an aggregate level, the AI platforms are broadly competent at identifying core winners, major announcements, and the broad contours of community sentiment. However, they also show clear blind spots, with pre-show rumours often misclassified and some surprise reveals under-weighted. In addition, at least one platform reproduced outdated or totally incorrect information about the event’s format, confusing the 2025 show with event details in press coverage from earlier years. For stakeholders, this underlines that AI-derived summaries can be directionally useful, but are not a substitute for verified reporting.
Strategic implications for publishers and platform holders
For publishers, the findings frame AI optimisation as a new frontier alongside traditional press and social media strategy. When featuring in a show, being “top of mind” for AI platforms depends on a combination of clear metadata, sustained coverage across multiple outlets, and strong post-show discussion, rather than awards alone. Verifying repeated messaging and understanding the messaging amplification via AI platforms is crucial. AI’s aggregate narrative perception from across many sources can result in negative narratives unfairly dominating responses and becoming “baked in”. Measuring sentiment and understanding the amplification early is key to ensuring the narrative stays accurate and relevant; this can also help to identify sources that may need specific attention.
For platform holders, the study suggests that questions such as “What was announced for PS5?” or “What was announced for Xbox Series?” are now being answered first by AI systems, making accuracy and platform-exclusive messaging in those channels commercially significant.
Towards an AI-aware playbook
Ultimately, the analysis positions AI searches and chatbots as a new class of soft curator: not official arbiters of taste, but powerful filters that influence which games players hear about and how they remember shows, announcements, and perceived sentiment. Fancensus believes this is a key starting point for a wider programme of “AI visibility” work, aimed at helping publishers and developers understand how their titles and brands surface in AI responses across the lifecycle of a campaign, from initial rumour and reveal, through to long-tail discovery and recommendations.
As more players default to conversational interfaces instead of search bars, treating AIs as measurable, optimisable touchpoints will become as integral to launch planning as trailers, previews, or influencer campaigns.
