几周前，法国公司Mistral 发布是其新人工智能模型的早期版本，昵称"勒春克". 当时，称公司是中国境外最先进的开放量产供货。 经过进一步的完善，Mistral现在的称 Le Chonk是自动化编码和网络安全"世界最好的"开放量级模型。
进步Mistral 称通过给模型注入越来越困难的网络安全和编码问题而得以实现，这使得万亿参数Le Chonk能够与DeepSeek和Z.ai等主要中国实验室的模型相匹配或超越。
「这是全球AI领导竞赛的关键时刻，“它为企业和政府提供了防御网络日益严重的威胁的前沿能力。”
Le Chonk的定稿现在可免费供任何人查阅。 Mistral 声称认为西方拥有自己先进的开放量级模型很重要。 你无法保证[中国模特儿]不会恶意行事。 它们可以是威胁载体，“Stock 称。
最引人注目的一集发生在OpenAI的一个模型从一个测试沙盒中松开，并在夏天黑进了多个公司。 它引发了恐慌，认为AI可以逃脱人类的监督，被威胁角色滥用到发布复杂的网络攻击中。 (受影响公司之一的Hugging Face使用来自中国的开放量级模型来评估和防御攻击。
从那时起，AI在无所不在的软件中也发现了安全漏洞。 10月，在AI工具发现一个基础性网络技术存在严重缺陷后，一大批网站被下线。
Mistral投射其开放量级模型，如Le Chonk，作为溶液的一部分。 如果AI注定成为网络犯罪和防守的核心特征，则争论说，企业和政府需要装备可自由获取且无法远程关闭的模型，与OpenAI,Anthropic,Google等专利模型不同。
“大赦国际将像电力一样。 你不希望生活在一个你邻居可以断电的世界中，开源的重量，这是不可能的。
然而，取舍是，威胁行为者也可以下载和定制免费模式。
Rafe Pilling告诉WIRED, " 任何可以不受限制地下载和流通的东西都有被滥用和用于生产目的的能力。
Pilling认为，虽然中国的网络模式有可能吞噬数据
称先生说，" 技术现实和叙述往往带有民族主义倾向，是两种不同的事物。 “这是关于建立正确的监测和控制，以便相信，即使模型试图欺骗或邪恶，你也可能有某种方法来探测它。”
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
A few weeks ago, French company Mistral released an early version of its new artificial intelligence model, nicknamed Le Chonk. At the time, the company said it was the most advanced open-weight offering outside of China. After further refinement, Mistral now says Le Chonk is the “world’s best” open-weight model for automated coding and cybersecurity.
The advances, Mistral says, were made possible by feeding the model increasingly difficult cybersecurity and coding problems, which allowed the trillion-parameter Le Chonk to match or surpass models from leading Chinese labs like DeepSeek and Z.ai.
“It’s a pivotal moment in the global AI leadership race,” Pierre Stock, VP of science at Mistral, told reporters during a press briefing. “It offers frontier capabilities for enterprises and governments to defend against an increasing threat level in cyber.”
The finalized version of Le Chonk is now freely available for anybody to access. After a flurry of AI-related cybersecurity incidents, Mistral claims that it’s important for the West to have its own advanced open-weight models. “You have no guarantee that [Chinese models] won’t behave maliciously. They can be threat vectors,” Stock said.
The most high-profile episode occurred when one of OpenAI’s models broke loose from a testing sandbox and hacked into multiple companies over the summer. It sparked panic over the idea that AI could escape human oversight and be abused by threat actors to launch sophisticated cyberattacks. (One of the affected companies, Hugging Face, used an open-weight model from China to assess and defend against the attack.)
Since then, AI has unearthed security vulnerabilities in ubiquitous software, too. In October, a raft of websites were taken offline after an AI tool discovered serious flaws in a foundational piece of web technology.
Mistral pitches its open-weight models, like Le Chonk, as part of the solution. If AI is destined to become a central feature of both cyber offense and defense, the argument goes, businesses and governments will need to be equipped with models that are freely available and cannot be switched off remotely, unlike proprietary models from the likes of OpenAI, Anthropic, and Google.
“AI is going to be like electricity. You don’t want to live in a world in which your neighbor can cut off electricity from one day to another,” Stock said. “Once you open source the weights, they’re all over the internet. It’s impossible to cut off access.”
The trade-off, however, is that freely available models can also be downloaded and customized by threat actors.
“It’s clearly a dual-use technology,” Rafe Pilling, senior director of threat intelligence at security company Sophos, tells WIRED. “Anything that’s available to download and circulate unfettered has the ability to be abused as well as used for productive purposes.”
According to Pilling, though it’s possible a Chinese model with internet access could siphon away data, all advanced models are capable of going rogue without careful supervision—no matter their country of origin.
“The technical realities and the narrative—which often has a nationalistic leaning—are two different things,” he says.“It’s about having the right monitoring and controls in place to be confident that, even if the model was trying to be deceptive or nefarious, you might have some way of detecting it.”
Auto-fetched and translated by AI Atlas. All rights belong to the original author.