广义而言，可以将我们有关[威胁](threat-modeling.md) 或者适用于大多数人的目标的建议分为这几类。 ==你可能关注其中零个、 一个、 几个、 或所有这些可能性==， 你应该使用的工具和服务取决于你的目标。 You may have specific threats outside these categories as well, which is perfectly fine! 重要的是要去了解您选择的这些工具的优缺点，因为也许任何工具都不能够保护您免受所有可以想象到的威胁。
Typically, a form of <span class="pg-red">:material-target-account: Targeted Attack</span> that centers around a vulnerability or exploit introduced into otherwise good software either directly or through a dependency from a third party.
其中一些威胁可能比其他威胁更重要，具体取决于您的关注点。 For example, a software developer with access to valuable or critical data may be primarily concerned with <span class="pg-viridian">:material-package-variant-closed-remove: Supply Chain Attacks</span> and <span class="pg-red">:material-target-account: Targeted Attacks</span>. They will likely still want to protect their personal data from being swept up in <span class="pg-blue">:material-eye-outline: Mass Surveillance</span> programs. 同样，"普通人 "可能主要关心他们的个人数据的 <span class="pg-green">:material-account-search: ，公开曝光</span> ，但他们仍应警惕那些侧重于安全的问题，比如<span class="pg-orange">:material-bug-outline: ，被动攻击</span>，就像那些会影响到设备的恶意软件 。
应用程序无法获得根访问权限，只能访问您授予它们访问权限的系统资源。 ChromeOS has similar sandboxing capabilities to Android, and macOS has full system permission control (and developers can opt in to sandboxing for applications). ChromeOS具有与安卓类似的沙盒属性，而macOS具有完整的系统权限控制和（针对开发者）可选的应用程序沙盒，然而这些操作系统的确会将识别信息传输给各自的OEM。 Linux倾向于不向系统供应商提交信息，但它对漏洞和恶意应用程序的保护很差。 This can be mitigated somewhat with specialized distributions which make significant use of virtual machines or containers, such as [Qubes OS](../desktop.md#qubes-os).
If you are concerned about **physical attacks** you should use an operating system with a secure verified boot implementation, such as Android, iOS, macOS, or [Windows (with TPM)](https://learn.microsoft.com/windows/security/information-protection/secure-the-windows-10-boot-process). 你还应该确保你的驱动器是加密的，并且操作系统使用TPM或安全 [Enclave](https://support.apple.com/guide/security/secure-enclave-sec59b0b31ff/1/web/1) 或 [Element](https://developers.google.com/android/security/android-ready-se) ，以限制输入加密口令的重试速率。你应该避免与你不信任的人分享你的电脑，因为大多数桌面操作系统没有按用户单独加密数据。
Supply chain attacks are frequently a form of <span class="pg-red">:material-target-account: Targeted Attack</span> towards businesses, governments, and activists, although they can end up compromising the public at large as well.
<div class="admonition example" markdown>
<p class="admonition-title">Example</p>
A notable example of this occurred in 2017 when M.E.Doc, a popular accounting software in Ukraine, was infected with the *NotPetya* virus, subsequently infecting people who downloaded that software with ransomware. NotPetya itself was a ransomware attack which impacted 2000+ companies in various countries, and was based on the *EternalBlue* exploit developed by the NSA to attack Windows computers over the network.
</div>
There are few ways in which this type of attack might be carried out:
1. A contributor or employee might first work their way into a position of power within a project or organization, and then abuse that position by adding malicious code.
2. A developer may be coerced by an outside party to add malicious code.
3. An individual or group might identify a third party software dependency (also known as a library) and work to infiltrate it with the above two methods, knowing that it will be used by "downstream" software developers.
These sorts of attacks can require a lot of time and preparation to perform and are risky because they can be detected, particularly in open source projects if they are popular and have outside interest. Unfortunately they're also one of the most dangerous as they are very hard to mitigate entirely. We would encourage readers to only use software which has a good reputation and makes an effort to reduce risk by:
1. Only adopting popular software that has been around for a while. The more interest in a project, the greater likelihood that external parties will notice malicious changes. A malicious actor will also need to spend more time gaining community trust with meaningful contributions.
2. Finding software which releases binaries with widely-used, trusted build infrastructure platforms, as opposed to developer workstations or self-hosted servers. Some systems like GitHub Actions let you inspect the build script that runs publicly for extra confidence. This lessens the likelihood that malware on a developer's machine could infect their packages, and gives confidence that the binaries produced are in fact produced correctly.
3. Looking for code signing on individual source code commits and releases, which creates an auditable trail of who did what. For example: Was the malicious code in the software repository? Which developer added it? Was it added during the build process?
4. Checking whether the source code has meaningful commit messages (such as [conventional commits](https://conventionalcommits.org)) which explain what each change is supposed to accomplish. Clear messages can make it easier for outsiders to the project to verify, audit, and find bugs.
5. Noting the number of contributors or maintainers a program has. A lone developer may be more susceptible to being coerced into adding malicious code by an external party, or to negligently enabling undesirable behavior. This may very well mean software developed by "Big Tech" has more scrutiny than a lone developer who doesn't answer to anyone.
On the other hand, web-based E2EE implementations, such as Proton Mail's web app or Bitwarden's *Web Vault*, rely on the server dynamically serving JavaScript code to the browser to handle cryptography. 一个恶意的服务器可以针对一个特定的用户，向他们发送恶意的JavaScript代码来窃取他们的加密密钥，而用户是很难注意到这样的事情的。即使用户注意到有人试图窃取他们的密钥，也很难证明是提供商试图这样做，因为服务器可以选择向不同的用户提供不同的网络客户端。
即使有端对端加密，服务提供商仍然可以根据 **元数据**，对你进行剖析，而这些元数据通常不受保护。 While the service provider can't read your messages, they can still observe important things, such as whom you're talking to, how often you message them, and when you're typically active. 对元数据的保护是相当不常见的，如果你关心这一点，应该密切关注你所使用的软件的技术文档，看看是否有任何元数据最小化或保护。
If you want to learn more about surveillance methods and how they're implemented in your city you can also take a look at the [Atlas of Surveillance](https://atlasofsurveillance.org) by the [Electronic Frontier Foundation](https://eff.org).
In France, you can take a look at the [Technopolice website](https://technopolice.fr/villes) maintained by the non-profit association La Quadrature du Net.
政府经常为大规模监控项目辩护，认为这是打击恐怖主义和防止犯罪的必要手段。 However, as breaches of human rights, they're most often used to disproportionately target minority groups and political dissidents, among others.
<p class="admonition-title">ACLU: <em><a href="https://aclu.org/news/national-security/the-privacy-lesson-of-9-11-mass-surveillance-is-not-the-way-forward">The Privacy Lesson of 9/11: Mass Surveillance is Not the Way Forward</a></em></p>
In the face of Edward Snowden's disclosures of government programs such as [PRISM](https://en.wikipedia.org/wiki/PRISM) and [Upstream](https://en.wikipedia.org/wiki/Upstream_collection), intelligence officials also admitted that the NSA had for years been secretly collecting records about virtually every American’s phone calls — who’s calling whom, when those calls are made, and how long they last. 你应该考虑你的对手能观察到网络的哪些方面，以及你的行动是否有合理的可否认性。
If you're concerned about mass surveillance programs, you can use strategies like compartmentalizing your online identities, blending in with other users, or, whenever possible, simply avoiding giving out identifying information.
Additionally, even companies outside the *AdTech* or tracking industry can share your information with [data brokers](https://en.wikipedia.org/wiki/Information_broker) (such as Cambridge Analytica, Experian, or Datalogix) or other parties. 例如，如果您的帐户具有“隐私模式” ，请启用此功能以确保您的帐户不会被搜索引擎索引，并且不会被未经您事先审核的人查看。对企业数据收集最有力的保护是尽可能地加密或混淆你的数据，使不同的供应商难以将数据相互关联并建立你的档案。
企业平台的审查制度越来越普遍，因为像Twitter和Facebook这样的平台屈服于公众需求、市场压力和政府机构的压力。 Government pressures can be covert requests to businesses, such as the White House [requesting the takedown](https://nytimes.com/2012/09/17/technology/on-the-web-a-fine-line-on-free-speech-across-globe.html) of a provocative YouTube video, or overt, such as the Chinese government requiring companies to adhere to a strict regime of censorship.
People concerned with the threat of censorship can use technologies like [Tor](../advanced/tor-overview.md) to circumvent it, and support censorship-resistant communication platforms like [Matrix](../social-networks.md#element), which doesn't have a centralized account authority that can close accounts arbitrarily.
[^4]: "[Enumerating badness](https://ranum.com/security/computer_security/editorials/dumb)" (or, "listing all the bad things that we know about"), as many content blockers and antivirus programs do, fails to adequately protect you from new and unknown threats because they have not yet been added to the filter list. 你还应该采用其他缓解技术。
