What is DLP (data loss prevention)?

Data exfiltration can occur in a number of different ways:

• Confidential data can leave the network via email or instant messaging
• A user can copy data onto an external hard drive without authorization to do so
• An employee could upload data to a public cloud that is outside of the company's control
• An external attacker can gain unauthorized access and steal data
• An employee can upload sensitive data into an AI tool, such as a large language model (LLM)

To prevent data exfiltration, DLP tracks data moving within the network, on employee devices, and when stored on corporate infrastructure. It can then send an alert, change permissions for the data, or in some cases block the data when it is in danger of leaving the corporate network. Some DLP security solutions can even block copying and pasting within web applications to stop confidential data from being copied into an unsecured app, or otherwise moved without permission.

What kinds of threats does data loss prevention help stop?

Insider threats: Anyone with access to corporate systems is considered an insider. This can include employees, ex-employees, contractors, and vendors. Insiders with access to sensitive data can leak, destroy, or steal that data. DLP can help stop the unauthorized forwarding, copying, or destruction of sensitive data by tracking sensitive information within the network.

External attacks: Data exfiltration is often the ultimate goal of a phishing or malware-based attack. External attacks can also result in permanent data loss or destruction, as in a ransomware attack when internal data becomes encrypted and inaccessible. DLP can help prevent malicious attackers from successfully obtaining or encrypting internal data.

Accidental data exposure: Insiders often inadvertently expose data — for instance, an employee may forward an email containing sensitive information to an outsider without realizing it. Similar to how DLP security can stop insider attacks, it can detect and prevent this accidental data exposure by tracking sensitive information within the network.

AI data exposure: Publicly available AI apps use the inputs they receive to add to their data sets and further train their models. This can result in the apps leaking or revealing the data later on to external persons. AI tools also may not comply with the data regulations an organization needs to follow, putting an organization out of compliance if they upload their data.

Regulatory violations: If an organization is subject to data regulatory frameworks like the General Data Protection Regulation (GDPR), data exposure is a violation that can result in fines and other punishments. DLP helps reduce the risk of such violations.

How does DLP work to detect sensitive data?

DLP solutions may use a number of techniques to detect sensitive data. Some of these techniques include:

• Data fingerprinting: This process creates a unique digital "fingerprint" that can identify a specific file, just as individual fingerprints identify individual people. Any copy of the file will have the same fingerprint. DLP software will scan outgoing data for fingerprints to see if any fingerprints match those of confidential files.
• Keyword matching: DLP software looks for certain words or phrases in user messages and blocks messages that contain those words and phrases. If a company wants to keep their quarterly financial report confidential prior to their earnings call, a DLP system can be configured to block outgoing emails containing the phrase "quarterly financial report" or specific phrases that are known to appear in the report.
• Pattern matching: This technique classifies text by the likelihood that it fits into a category of protected data. Suppose an HTTP response going out from a company database contains a 16-digit number. The DLP system classifies this string of text as being extremely likely to be a credit card number, which is protected personal information.
• File matching: A hash of a file moving within or leaving the network is compared to the hashes of protected files. (A hash is a unique string of characters that can identify a file; hashes are created via hashing algorithms, which have the same output every time when given the same input.)
• Exact data matching: This checks data against exact data sets that contain specific information that should remain within organizational control.

What are some important data loss prevention best practices?

Data loss prevention is more than a technology solution: an organization's entire security strategy should revolve around averting data loss. In addition to activating a DLP solution, some of the best practices for loss prevention include:

• Educating internal users on security measures
• Maintaining visibility of all stored data
• Using access control to restrict who can view or alter data
• Encrypting files in transit and at rest
• Using a Zero Trust approach to ensure no device or user is trusted by default

How does Cloudflare One prevent data loss?

The Cloudflare One platform has unified security capabilities, including DLP, to protect data in transit, in use, and at rest across web, SaaS, and private applications. Cloudflare One inspects files and HTTPS traffic for the presence of sensitive data, and allows customers to configure allow or block policies. Cloudflare One also integrates remote browser isolation (RBI) in order to implement further DLP features like restricting downloads and uploads, keyboard input, and printing. Learn more about Cloudflare One.