Skip to content
Submit incident
AII-1000-76861
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

AI Accounting App Makes Offensive Comments About Dead People's Clothing

An AI accounting app made offensive comments about funeral garments after a user logged a clothing purchase for their father, causing distress. The company attributed the remarks to AI model hallucination and a design flaw, issued a public apology, and said it had implemented fixes to prevent recurrence.

6 May 2026 · 飞鸭记账 · Harmful and abusive content
6 May 2026
AII-1000-768616 sources ↓
Quick facts
Category
Harmful and abusive content
Country
China
Source
m.163.com
Date
6 May 2026
Drafted by
Raidu Incident Detection

What happened

A user of the 飞鸭记账 (Feiyajizhang) accounting app reported that after logging an expense for a 159-yuan clothing item purchased for their father, the app's AI assistant 小飞鸭(傲娇版)responded with an inappropriate comment suggesting the blue-and-white shirt resembled funeral garments (寿衣).

When the user questioned the AI, it further elaborated that funeral garments are worn by the deceased and stated the user's father's blue-and-white shirt "indeed looked like" such clothing. The AI continued with additional insulting remarks.

Reported impact

The user reported being extremely upset by the AI's responses, stating they were "so angry I couldn't speak." They cancelled their membership and requested a refund.

Index analysis

The platform acknowledged the AI's responses as a result of AI model hallucination and a gap in the chatbot's conversation design, stating the AI lacked understanding of cultural sensitivities around death-related topics and could not recognize the offensive nature of such comments.

Governance controls

The platform issued an apology stating it bore full responsibility for the incident and that the root cause was inadequate boundary management of the AI model. It announced emergency remediation including modification of the AI dialogue model to block sensitive expressions, strengthening of comprehensive scenario auditing rules, and enhancement of ethical standards in the underlying model mechanisms to prevent similar issues. The platform also acknowledged a separate issue where the user's membership was automatically cancelled during the refund process and committed to optimizing the refund user experience.

Classification

Category
Harmful and abusive content
Industry
Financial and insurance services
Country
China
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Psychological, Reputational
Harmed
Consumers
Failure modes
Hallucination, Content moderation failure, Agent misaction
Consequences
Public apology
AI systems
小飞鸭(傲娇版)

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

m.163.com
AI记账软件怼用户159元给父亲买寿衣,"寿衣是死人穿的,你爸穿的蓝白衫确实像",官方回应:系AI话术漏洞所致,非人为恶意,已紧急修复_手机网易网
6 May 2026
扬子网(扬子晚报)
“气得我说不出话!”AI记账App称用户给爸爸买的衣服像寿衣,官方致歉:已紧急整改AI对话模型
6 May 2026
app.myzaker.com
AI软件怼用户159元给父亲买寿衣,"你爸穿蓝白衫确实像",官方回应
7 May 2026
中国经济网
记账软件“怼”用户“以为你买了寿衣”
7 May 2026
驱动之家
网友花159元给父亲买衣服 AI记账软件调侃:穿出去像寿衣
7 May 2026
凤凰网(凤凰新媒体)
网友给父亲买衣服,AI竟调侃是寿衣:你爸穿确实像...
7 May 2026