Tiếp theoHướng dẫn tiếp theo
Viết chính sách sử dụng chấp nhận được của AI tại nơi làm việc
xã hội
HƯỚNG DẪN xã hội
Shadow AI is employees' use of AI tools, such as personal chatbot accounts, browser extensions or AI features in unapproved apps, without their organization's knowledge or approval.
Nghe hướng dẫn nàyAI thuật lại · ~3 min
It matters because confidential data can leave the company through prompts and uploads, and compliance obligations can be broken without anyone noticing. Outright bans tend to push use further underground rather than stopping it.
Shadow AI grows out of a simple gap: AI tools are useful, cheap and a browser tab away, while approved alternatives are often missing or less capable. Most employees who use unapproved AI are not being malicious. They are trying to write faster, analyze data or fix code. The term follows 'shadow IT', which describes unsanctioned software and cloud services, but AI raises the stakes because the whole point is to paste in real work content. The risks fall into a few groups. The first is data leakage. Prompts and uploaded files may be stored by the provider and, depending on the service and settings, used to improve models. Consumer and business tiers often differ, and business offerings from major providers generally exclude customer data from training by default. The best-known case is Samsung. In 2023 engineers reportedly entered confidential source code and meeting notes into ChatGPT, and Samsung then restricted generative AI use on company devices. The second is compliance. Sharing personal data with a processor that has no contract can breach privacy laws such as the GDPR, and regulated sectors have extra rules. The third is quality and accountability. Unreviewed AI output can end up in client work with no record of its source. The fourth is the tools themselves, such as extensions and plugins that ask for broad permissions. Organizations that manage shadow AI well tend to use a mix of measures instead of a ban. They offer approved enterprise tools with suitable data terms. They write a short acceptable-use policy that sorts data into what can and cannot go into AI tools. They use technical controls such as data loss prevention and monitoring of AI domains. And they give people a quick way to request new tools. A common misconception is that blocking one popular chatbot solves the problem. Hundreds of AI services and embedded features exist, and staff can use personal phones.
An engineer pastes proprietary source code into a personal chatbot account to find a bug. This is similar to reported 2023 incidents at Samsung that led the company to restrict generative AI tools.
A recruiter uploads a spreadsheet of candidate names, salaries and interview notes to a free AI tool to summarize it, sending personal data to a provider with no data processing agreement in place.
A marketing coordinator installs an AI writing browser extension that can read the content of every web page she opens, including internal dashboards.
A hospital administrator uses an unapproved transcription app to summarize a meeting where patient cases were discussed, which creates health-privacy compliance risk.
As AI features ship inside ordinary software such as office suites, browsers and operating systems, the line between approved and shadow AI will blur. The question becomes which AI features are turned on and under what data terms, not which websites employees visit. Autonomous agents that can act on data raise the stakes further. Organizations are likely to move toward AI governance programs that inventory tools, classify data and train staff, rather than rely on network blocks. Regulations such as the EU AI Act add obligations for some uses, which increases the need to know what AI is actually in use.
Bản tin miễn phí
Ba câu chuyện AI đã được xác minh vào mỗi buổi sáng các ngày trong tuần, được viết bằng tiếng Anh đơn giản. Miễn phí mãi mãi, không có quảng cáo.
Một email mỗi ngày trong tuần. Hủy đăng ký trong một cú nhấp chuột. Chúng tôi không bao giờ bán hoặc chia sẻ địa chỉ của bạn.
Tự kiểm tra
Phản hồi tức thì về mọi câu trả lời và chứng chỉ có thể chia sẻ với ID có thể xác minh sau khi bạn vượt qua khóa học.
Hỗ trợ giáo dục AI miễn phí. AI Understanding là tổ chức phi lợi nhuận 501(c)(3) — không bao giờ có quảng cáo, không có tường phí. Hãy quyên góp
Shadow AI is employees' use of AI tools, such as personal chatbot accounts, browser extensions or AI features in unapproved apps, without their organization's knowledge or approval. It matters because confidential data can leave the company through prompts and uploads, and compliance obligations can be broken without anyone noticing. Outright bans tend to push use further underground rather than stopping it.
Shadow AI is unsanctioned AI use, such as personal chatbot accounts, extensions or unapproved AI features.
The value of AI comes from feeding it real work content, which directly creates data-exposure risk.
Confidential source code and meeting notes were reportedly entered into ChatGPT, which led to restrictions on company devices.
Business tiers generally exclude customer data from training by default, which is one reason approved enterprise tools lower the risk.
Blocking one chatbot leaves many alternatives, and staff can switch to personal devices, so use continues unseen.
Tiếp tục học hỏi
Đã chọn thêm hướng dẫn cho chủ đề này
Tiếp theoHướng dẫn tiếp theo
Viết chính sách sử dụng chấp nhận được của AI tại nơi làm việc
xã hội