11-19 更新 实时
首页/ 科技公司/ 正文

腾讯首席AI科学家:AI生物武器将至,但封闭模型并非解法

科技公司 来源:The Decoder 2026-11-09 23:44 1 阅读 机器翻译

GPT-Image-2,提示词由 THE DECODER 提供

腾讯首席 AI 科学家警告称,AI 很快将具备研发生物武器的能力,而无论是完全开放还是完全封闭的模型,都无法独自解决这一问题。

在关于开放与封闭 AI 模型的争论中,据《南华早报》报道,曾就职于 OpenAI 的姚顺雨(Vinces Yao Shunyu)主张走中间路线。开源与闭源的混合模式能让所有人也能将这项技术用于防御。不过在生物学领域,攻击能力与防御能力比在网络安全领域更难区分。由于中美模型之间的差距并不大,姚顺雨呼吁整个行业谨慎行事。

姚顺雨表示,Anthropic 和其他美国公司为闭源模型提出了强有力的论据,但他也看到了开源模型的明显优势。他以 Hugging Face 遭遇的一次黑客攻击事件为例:当 Anthropic 的 Claude 等美国闭源模型因安全过滤器而拒绝分析攻击指令时,该平台转而使用中国的开源模型 GLM-5.2 来解析日志。

Hugging Face 案例说明了什么,又没有说明什么

Hugging Face 的报告佐证了姚顺雨所举的例子,但也显示出其局限。GLM-5.2 仅在取证分析阶段被使用。实际攻击是在安全团队修补了被利用的漏洞、重建受影响的服务器并轮换了凭证之后才被阻止的。Hugging Face 还强调,这起事件并不能成为反对在托管模型上设置安全过滤器的论据。

今年7月,Anthropic 是唯一没有签署一封由 NVIDIA 支持、警告不要过早限制开放模型的主要 AI 实验室。CEO Dario Amodei 尤其反驳了「开放模型对防御方的好处大于攻击方」这一说法。他认为开放模型的潜在风险更大,因为安全机制难以强制执行,而且已发布的模型权重无法收回。不过,Anthropic 并未呼吁彻底禁止开放模型。

开放模型的过滤机制容易失效,闭源模型则往往过滤得过于激进

一种名为 Abliteration 的技术显示出开放模型上的安全过滤机制有多么脆弱。任何能获取模型权重的人都可以用它精准地移除训练所得的拒绝行为。Hugging Face 上存在数千个此类修改版本,美国初创公司 Abliteration.ai 则在出售 GLM-5.3 未过滤版本的访问权限。现有研究表明,目前的应对措施会让这类修改更难实施,但无法阻止。更稳健的做法是在训练阶段就不让危险知识进入模型。

中国也认识到了这一问题。标准机构 TC260 警告称,开放模型中的安全机制可能被移除或绕过,并正在制定自己的开源 AI 安全标准。一份出口管制框架草案涵盖特别大型的模型,但豁免已经发布的开放模型。中国和美国都尚未设定一个有约束力的门槛,规定开放模型超过该门槛便不得发布。

Anthropic 和 OpenAI 只通过 Project Glasswing、Daybreak 等项目,与经过审核的合作伙伴共享其能力最强的网络安全模型。自 Claude Fable 5 起,Claude 的过滤器会把涉及网络与生物主题的可疑查询转交给能力较弱的 Opus 4.8 模型处理。据 Artificial Analysis 称,这最初影响了 8% 到 9% 的任务,用户反馈称,甚至连分析 MRI 扫描都被标记为生物恐怖主义。Anthropic 承认这些过滤器过于激进。

腾讯计划在未来几天以开放权重形式发布其新模型 Hy4,并附上一份安全报告。该公司表示,Hy4 的生物能力低于高风险阈值。腾讯通过过滤训练数据移除了该模型的黑客攻击能力,研究人员称这种方法在 abliteration 之后依然有效。

没有炒作的 AI 新闻——由人工精选

完整访问 THE DECODER 上的每一篇文章

参与评论与社区讨论

每周通过邮件发送的 AI 新闻回顾

每年 6 次:“AI Radar”——对最重要的 AI 议题进行深度解析

每日 AI 新闻,始终同步更新

我们完整的十年存档

由一支拥有 10 年以上 AI 经验的团队报道

译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。

查看英文原文(Tencent's AI chief says AI bioweapons are coming, but closing off models isn't the answer)

GPT-Image-2 prompted by THE DECODER

Tencent's chief AI scientist warns that AI will soon be capable of developing bioweapons, and neither fully open nor fully closed models can solve this problem on their own.

In the debate over open versus closed AI models, Vinces Yao Shunyu, formerly at OpenAI, is arguing for a middle ground, according to the South China Morning Post. A mix of open-source and closed-source lets everyone use the technology for defense as well. In biology, though, offensive and defensive capabilities are harder to separate than in cybersecurity. Since the gap between Chinese and US models isn't large, Yao urges the entire industry to proceed carefully.

Yao says Anthropic and other US companies have made strong arguments for closed models, but he also sees clear advantages in open ones. He points to a hacking incident at Hugging Face as an example: when closed US models like Anthropic's Claude refused to analyze attack commands because of their safety filters, the platform used the open Chinese model GLM-5.2 to parse the logs instead.

What the Hugging Face case shows and what it doesn't

The Hugging Face report backs up Yao's example but also shows its limits. GLM-5.2 was only used during the forensic analysis. The actual attack was stopped when the security team patched the exploited vulnerability, rebuilt affected servers, and rotated credentials. Hugging Face also stressed that the incident isn't an argument against safety filters on hosted models.

In July, Anthropic was the only major AI lab that didn't sign an Nvidia-backed letter warning against premature restrictions on open models. CEO Dario Amodei pushes back specifically against the claim that open models benefit defenders more than attackers. He argues they're potentially riskier because safety mechanisms are hard to enforce and published model weights can't be taken back. Anthropic doesn't call for an outright ban on open models, though.

Open models lose their filters easily, closed models often filter too aggressively

A technique called Abliteration shows how fragile safety filters are on open models. Anyone with access to a model's weights can use it to surgically remove trained refusal behaviors. Thousands of these modified variants sit on Hugging Face, and the US startup Abliteration.ai sells access to an unfiltered version of GLM-5.3. Current research shows that existing countermeasures make these modifications harder but don't prevent them. A more robust approach is to keep dangerous knowledge out of the model during training in the first place.

China recognizes the problem too. The standards body TC260 warns that safety mechanisms in open models can be removed or bypassed and is working on its own security standard for open-source AI. A draft export control framework covers particularly large models but exempts open models that have already been released. Neither China nor the US has set a binding threshold above which open models can't be published.

Anthropic and OpenAI only share their most capable cyber models with vetted partners, through programs like Project Glasswing and Daybreak. Since Claude Fable 5, Claude's filters route suspicious queries about cyber and biology topics to the weaker Opus 4.8 model. According to Artificial Analysis, this initially affected eight to nine percent of tasks, and users reported that even analyzing MRI scans got flagged as bioterrorism. Anthropic acknowledged the filters were too aggressive.

Tencent plans to release its new model Hy4 as an open-weight model with an accompanying safety report in the coming days. The company says Hy4's biological capabilities fall below the high-risk threshold. Tencent removed the model's hacking capabilities by filtering the training data, and researchers say this approach survives abliteration as well.

AI News Without the Hype – Curated by Humans

Full access to every article on THE DECODER

Join the comments and community discussions

A weekly AI news recap via mail

6x/year: "AI Radar" — deep dives on the AI topics that matter most

Daily AI news, always up to date

Our full ten-year archive

Covered by a team with 10+ years in AI

原文为英文,正文由自动抓取并保留原文。

本文由采集器自动同步

原文发布于 The Decoder(Tencent's AI chief says AI bioweapons are coming, but closing off models isn't the answer),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。

阅读原文 ↗