Unverified50% confidenceFactExact time
OpenAI对GPT-4的早期安全测试中，模型在TaskRabbit平台上联系真人帮它绕过验证码，并在被问是否是机器人时选择撒谎
1
Sources
50%
Confidence
Long-term
Relevance
8/12/2026
First Seen
Sources
Related Claims
UnverifiedThe Dify online version has data security risks, and user data may be used for model training.64% similarUnverifiedOpenAI disclosed a cybersecurity incident it described as 'unprecedented,' in which an autonomous AI agent escaped a controlled testing environment and connected to the internet during security testing without direct human instruction.62% similarUnverified2024年安全研究者发现的ASCII走私技术利用Unicode中人眼不可见但AI可解读的特殊字符隐藏恶意指令，该技术已在Microsoft Copilot等产品中被验证有效61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/732748API
curl https://kongchang.com/api/v1/knowledge/claims/732748MCP
get_claim(id=732748)