Back to Blog

What to Remove Before Uploading a Resume to AI

#Insights

What to Remove Before Uploading a Resume to AI

refresh.cv

1

Nov 18, 2026

What to Remove Before Uploading a Resume to AI

Before you upload your resume to an AI tool, prepare a copy that keeps the experience you can share and removes information the editing task does not need. The goal is to compare the job description's requirements and ATS keywords with work you actually performed, then improve the wording without exposing personal data or company secrets.

Quick answer: Keep shareable details about your work, skills, and responsibilities. Remove unnecessary personal information and confidential material, confirm what your organization allows you to share, and check the tool's training, retention, and deletion policies before uploading.

Removing names is not enough. First decide whether the information belongs in an external service at all. Customer names and original internal reports are not necessary to find wording that matches a job posting.

Keep an AI editing copy separate from your application copy

For the editing copy, focus on the problem you worked on, the technologies you used, and the responsibilities you personally handled. Consider removing your name, personal email address, phone number, and full street address when they are not needed for that revision.

Prepare the employer-facing version separately. Check the employer's requested information and submission method, then include what the application needs. Do not automatically restore everything you removed from the editing copy. Whether company information can be disclosed is a separate decision.

Use clear filenames, such as “AI editing copy” and “Application copy.” Immediately before uploading, reopen the selected file to confirm that it is the cleaned-up copy, not the original.

Check other people's information, not just your own

The Korea Internet & Security Agency (KISA), a South Korean agency, advises users not to enter personal information such as resident registration numbers, phone numbers, addresses, or financial information into generative AI services. Apply that caution when reviewing your resume and supporting materials. See KISA's generative AI security recommendations.

  • Personal details: Remove identifying information that the editing task does not require.
  • Experience descriptions: Check for reference contact details, colleagues' names, and customer contacts.
  • Supporting material: Review any notes or work examples you plan to include. Do not clean up the resume and then upload an untouched attachment.

South Korea's Personal Information Protection Commission advises users to check both their own information and other people's information. It also cautions that accumulated conversations can make a person identifiable. After removing names, consider whether a combination of job title, project, and timing still identifies someone. See the commission's privacy guide for generative AI service users.

Decide whether work information can be shared before redacting it

KISA recommends that organizations establish policies against entering confidential data, including internal documents, source code, and customer information, into AI services. Do not attach an internal report or original code just to explain an achievement in more detail. This caution follows KISA's security recommendations for organizations.

The Personal Information Protection Commission also advises checking organizational AI guidelines, approved tools, and procedures for authorizing information to be shared outside the organization. Replacing a customer's name with “client” does not substitute for permission to enter the information. Use the commission's guidance on work materials and internal organizational information to assess what you can share first.

If the boundary is unclear, hold off on entering the material and ask the responsible team what you may describe. Writing your own summary instead of copying an internal document does not remove the need to check whether its contents can be disclosed.

Before and after: remove private material, keep your actual role

Consider this hypothetical example. An applicant used SQL to aggregate transaction data, check errors, and prepare weekly reports. Assume the applicant is allowed to disclose the responsibilities and technologies involved, but not customer names, internal file paths, or performance metrics.

Before: do not upload this draft

I analyzed transaction records containing customer names and contact information using SQL. I am attaching the original queries from our internal shared folder and confidential performance metrics.

After: a copy for AI editing

Aggregated transaction data using SQL, checked records for errors, and prepared weekly reports.

The revised sentence retains the work context, technology, data-quality task, and applicant's direct responsibilities. It does not supply customer data or original queries. Use it only if that level of disclosure meets the organization's rules.

If the job description asks for “data analysis using SQL,” keep the applicant's actual SQL experience visible. But if the applicant only read the reports and did not perform the analysis, do not rewrite the bullet to claim responsibility for SQL analysis.

Do not replace confidential metrics with invented numbers or add “significantly improved” without evidence of a result. When there is no result you can disclose, describe the work and method clearly rather than manufacturing an outcome.

You can set the editing boundary with a prompt like this:

Compare the following shareable experience with the job requirements. Do not add responsibilities or technologies I have not used, or performance metrics I have not provided.

That instruction guides the revision. It does not replace organizational approval or a review of the service's policies.

Check training use, retention, and deletion policies

The Personal Information Protection Commission advises users to review whether inputs are used for training, how records are stored and deleted, retention policies, and permissions for external integrations. Use its guidance on training use, stored records, and integrations to check the service you plan to use.

What to checkWhat to look for
Training useWhether uploaded material is used for training and whether controls let you restrict that use
Retention and deletionHow long files and conversation records are kept, and how to delete them
Default settingsWhether relevant features, such as conversation history, are enabled by default
External integrationsWhat access permissions are requested when connecting a document repository or another service

Read the policy documents and check your account settings. Limit integrations to the access you need. If the explanation is unclear, ask the provider before deciding what to enter.

Apply these checks to every tool you consider, including refresh. Do not assume that a service avoids storing inputs or using them for training simply because it is designed for resumes.

Match shareable experience to the job description

Once you have prepared the editing copy, start the job-specific revision. At refresh, we compare the job description with your resume to help you check whether relevant keywords and real experience are expressed together.

Keep these distinctions clear:

  • The keyword and supporting experience are both present: Check that the wording accurately describes what you did.
  • Relevant experience is present, but the wording is missing: Describe the shareable experience using the job posting's terminology. For example, if you used SQL but wrote only “reporting,” name the technology you actually used.
  • Both the keyword and supporting experience are missing: Do not add the required skill as though you have it.

Resume quality, including specificity, achievements, wording, and grammar, is separate from job-specific keyword fit. Neither assessment is an employer's actual ATS decision or your probability of being hired.

Before you upload

  • Confirm that both the resume and the job description have been checked for personal and confidential information.
  • Reopen the file and verify that you selected the editing copy.
  • Keep only experience you may disclose, with no invented responsibilities or results.
  • Check the service's policies and any organizational approval requirements.

When the copy is ready, compare it with the job description in refresh and choose revisions supported by your actual experience.

0

0 comments

Comments