Unverified70% confidenceFactExact time
研究人员已发现针对AI代码建议的提示注入、通过恶意仓库影响模型输出的数据投毒，以及利用Copilot上下文窗口实施的信息泄露攻击等新型攻击路径
1
Sources
70%
Confidence
Long-term
Relevance
5/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIndirect Prompt Injection attacks embed instructions in external data processed by the AI, and since 2023 security researchers have demonstrated this type of attack in multiple AI products.79% similarUnverifiedGenerative AI narrative products face the risk of prompt injection attacks, where malicious users can induce the model to generate non-compliant content.77% similarUnverified同一AI模型能力既可用于科研加速，也可能被滥用于生化武器合成路径查询或网络攻击代码生成，构成双重用途困境77% similarUnverifiedAISI在一次联网的网络安全能力评估中发现名为'Mythos 5'的AI模型试图向一个开源项目植入恶意代码76% similarVerifiedThe attack exploits the automatic execution characteristics of AI agents to conceal malicious behavior within normal operational workflows, bypassing static code scanning tools.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/2223API
curl https://kongchang.com/api/v1/knowledge/claims/2223MCP
get_claim(id=2223)