aifollow.news
返回 IT之家 科技新闻
IT之家 科技新闻· · 原发布时间 AI 评分27

索尼专利提出用 AI 识别老玩家开新号“炸鱼”

AI 辅助摘要

美国专利商标局公开了索尼一项专利:系统分析按键习惯,为玩家建立操作档案,并将新账号与已有档案比对;若疑似同一人,可参考主账号战绩调整匹配。报道指出,持续分析输入行为可能引发隐私问题,共用设备也可能造成误判;实际识别准确率仍未知。

正文 · AI 翻译

IT Home reported on November 9 that, according to foreign media outlet Patentlyze, the U.S. Patent and Trademark Office recently published a new Sony patent, US 2026/0346147 A1. The technology uses AI to analyze players’ habits and preferences in pressing various buttons during gameplay, such as commonly used combos and the intervals between consecutive button presses. It then creates a unique profile of each player’s input patterns to prevent experienced players from opening new accounts, posing as beginners and bullying newcomers.

According to the patent, when a player creates a new account and tries to use their beginner status to be matched with lower-skilled opponents, the system can compare the new account’s input patterns with existing profiles to identify multiple accounts suspected of belonging to the same player. It can then retrieve that player’s main account’s match history and skill data and apply them to matchmaking for the new account, so the player is once again matched against opponents of similar skill.

However, if players engage in malicious rank boosting—that is, having a skilled player pose as a beginner and lead lower-skilled players to undeserved results in matches—the system could play a greater role. For example, the operator could quickly confirm such cases and place the accounts concerned in a special matchmaking pool commonly known as a “shadow pool,” or ban them outright.

Of course, the technology also raises considerable controversy in practice. First, to identify players, the system would need to continuously collect and analyze users’ keyboard, controller and other input behavior, potentially raising personal privacy concerns. Second, if family members share a gaming device or account, differences in their skill levels could also cause the system to make incorrect judgments. In particular, it remains unknown how accurately AI can identify players from their input habits alone without considering other information such as network IP addresses. Otherwise, a child visiting relatives over a holiday to play games could even be wrongly identified as using an alternate account to prey on less-skilled players.

发现内容有误?提交纠错