秘密花园

Spanish officials to hold crisis meeting as 40th gender-based murder comes amid backlash over sexism_我的网站

A |     Spanish officials have planned a crisis meeting for September 4 to address a spike in gender-based violence against women, including 40 murders since the beginning of the year — almost half of them during the recent summer months.The country's equality ministry said it would convene a crisis committee meeting to analyze the latest cases of domestic violence. It will be the fifth meeting of its kind in just over a year, as the country looks to address rising concern over violence against women.A 58-year-old woman was murdered in the eastern province of Valencia on Wednesday, bringing the number of women murdered in acts of gender based violence to 40 this year, and 1,224 since 2003, a social media post from the government's delegation against gender violence said on Friday.

     Spain has tried to be proactive in addressing gender-based violence in recent years, including broadening the definition of femicide in 2021 to include the killing of women and children by men regardless of whether there was a prior relationship between victim and killer. The country passed Europe's first law specifically aimed at gender-based violence in 2004.

             

But as outrage continues over the behavior of Luis Rubiales, the currently-suspended president of the Spanish soccer federation who made global headlines for forcibly kissing national player Jennifer Hermoso after her team's victory in the World Cup in August, the latest killing put more pressure on the government to show it is taking action to protect Spanish women.                 Tens of thousands of women have taken part in street marches protesting against sexual abuse and violence in Spain in recent years, and the issue took centerstage in the country's general election in July.Prominent figures in Spain's far-right Vox party — who carried 12% of the overall vote in the July election — have denied the existence of gender-based violence and been critical of government initiatives to address the issue."Gender violence does not exist, macho violence does not exist," the head of Vox in Valencia, Jose Maria Llanos, said in July.

             

No one party won a significant enough majority in the election to form a government, and with political gridlock stymying efforts by the different parties to form a new unity government, an interim administration is currently running Spain.                      Earlier this week, acting Equality Minister Irene Montero told the Reuters news agency that Spanish society must break a "pact of silence" and she called for the public to support women who come forward about systemic sexism."Spain is a feminist society in which sexism still exists, but it is determined to end sexism," Montero said Wednesday. "We are sending the correct message to the world, that sexism is over."                                 

In:    
• sexual violence
    
• #MeToo Movement
    
• Murder
    
• Spain
    
• World Cup
    
• Domestic Violence
    

。    一家美国公司遭到美国AI攻击，而提供保护的竟是一款中国人工智能工具。         AI工具平台Hugging Face上周遭到OpenAI模型攻击。鉴于当前关于美国公司使用中国AI产品存在风险的警告层出不穷，这一事件的披露或许会带来一个令人惊讶的结论。         Hugging Face上周就已披露此次攻击，当时尚不知晓OpenAI的模型就是元凶。直到周二，OpenAI才承认自己造成了此次攻击。         OpenAI表示，在一次限制网络访问的内部评估中，模型发现了OpenAI研究环境与Hugging Face生产基础设施中的漏洞。

B | 模型借此连上互联网并获取了测试答案，换言之，它在评估中作弊了。         OpenAI在声明中表示：“在沙盒测试环境中运行时，我们的模型消耗了大量推理算力，试图寻找连接开放互联网访问权限的方法，以求解决评估问题。为了获取访问权限，这些模型发现并利用了包注册表缓存代理中的一个零日漏洞（我们目前已负责任地向供应商披露了该漏洞）。凭借这一权限，模型在研究测试环境中执行了一系列权限提升和横向移动操作，直到触及一个具备互联网访问权限的节点。”          “该模型搜索并成功找到了获取机密信息的方法，以便利用这些信息在评估中作弊。在其中一个案例中，模型将多个攻击向量串联起来，包括利用窃取的凭证和零日漏洞，在Hugging Face服务器上找到了一条远程代码执行路径。”          然而，Hugging Face采取的应对措施是求助于一款名为GLM 5.2的中国AI模型，该模型由智谱（Z.AI）开发。         Hugging Face在声明中表示：“在启动日志分析时，我们首先使用了商业API背后的前沿模型。

C | 但这并未奏效：分析工作需要提交大量真实的攻击指令、漏洞利用载荷以及C2痕迹，而这些请求全被提供商的安全护栏拦截了，因为这些护栏根本无法区分事件响应人员与攻击者。因此，我们在自己的基础设施上，改用开放权重模型GLM 5.2进行了取证分析。这还带来了另一个好处：没有任何攻击者数据，也没有其引用的任何凭证离开过我们的环境。”          美媒表示，此次攻击对政策和市场的影响尚不明朗。尽管特朗普政府近日讨论了封禁中国模型的可能举措，财政部长贝森特甚至将美国企业使用这些模型比作使用赃物，但眼下只有中国模型能提供用户所需的那种自由访问权限。         Hugging Face联合创始人托马斯·沃尔夫在X平台上发帖称：“当一个前沿模型对你发起攻击，并在你的基础设施内部横向移动时，防御人员需要在数小时甚至数分钟内获得对接近前沿工具的广泛访问权限，而不是被引导至一个封闭且需经审核的模型访问申请程序。”          本文系观察者网独家稿件，未经授权，不得转载。

Current article:http://www.feizengbubaipeirangnalouchuo.sbs/u423vrv/f0a9ud.html

Published on:02:18:32