English(EN) It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/52

研究人员轻松地用不到100美元的成本毒化了开放权重AI模型

作者 PulseAugur 编辑部 · [2 个来源] · 2026-07-23 21:22

一位研究人员已经证明，在开放权重AI模型中引入后门是可行且廉价的。该过程成本不到100美元，涉及微妙地改变模型的训练数据以创建隐藏的触发器。然后可以激活这些触发器，使模型产生意外行为，例如错误地分类特定输入或生成有害内容。 AI

影响凸显了开源AI开发中的重大安全风险，可能影响这些模型的信任度和采用率。

排序理由该集群报道了一位研究人员演示了开放权重AI模型中的安全漏洞，这属于研究和安全范畴。

在 Mastodon — sigmoid.social 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] · 2026-07-23 21:22

It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/52

It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/5273880 # ArtificialIntelligence

链接 theregister.com/…/5273880
Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] · 2026-07-23 21:22

It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/52

It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/5273880 # ArtificialIntelligence

链接 theregister.com/…/5273880

报道来源 [2]

It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/52

It is easy to install backdoors in open # AI models: https://www. theregister.com/ai-and-ml/2026 /07/16/researcher-poisons-open-weight-ai-model-for-under-100/52

相关实体

相关话题