Data Subject Access Request

What is a data subject access request?

A data subject access request (DSAR) is a person’s request to learn whether an organization processes their personal data and to get a copy. Under the EU General Data Protection Regulation (GDPR), it is the right of access in Article 15. US state laws such as the California Consumer Privacy Act (CCPA) grant a similar right to know.

For example, a former customer emails a bank asking for everything it holds about her. The privacy team pulls her account records and email history in a day. Then they find she also chatted with the bank’s AI assistant for months. Her name, address and complaints sit in prompt logs, in a vector database built from support transcripts and in an AI agent’s memory. Those copies count too. This is a summary, not legal advice.

What a DSAR covers under GDPR

Under Article 15, the person can get confirmation that their data is processed, a copy of that data and information about:

  • The purposes of processing and the categories of data.
  • The recipients, including any in other countries.
  • How long the data will be kept, or how that period is set.
  • Where the data came from, if not from the person.
  • Any automated decision-making, including profiling, with meaningful information about the logic involved.
  • Their rights to correction, erasure, restriction and objection, and to complain to a supervisory authority.

DSAR deadlines

Law Deadline Extension
GDPR (EU) One month from receipt Up to two more months for complex or numerous requests, if the person is told within the first month
CCPA (California) 45 calendar days One extension of 45 more days, with notice to the consumer

Under the GDPR, the first copy is free. An organization can charge a reasonable fee or refuse only when a request is manifestly unfounded or excessive. Under the CCPA, a business must confirm receipt within 10 business days and verify the requester’s identity. The right to know covers the prior 12 months, and data collected on or after January 1, 2022 unless providing it proves impossible or would take disproportionate effort.

Why DSARs are harder with AI systems

Traditional DSAR tools search databases and file shares by name or customer ID. AI systems spread personal data into places those tools do not index: prompts and chat histories, model provider logs, embeddings in vector stores, AI agent memory and the inputs and outputs of tool calls. Free text makes it worse, since a person may appear as a nickname, an email address or a phone number. A team that cannot find these copies cannot answer a DSAR fully, or honor a deletion request that follows it.

How Protecto helps

Protecto replaces personal data with consistent tokens before it enters prompts, logs and vector stores, and the same value always gets the same token. To answer a DSAR, a team can search masked prompts, logs and vector stores for that person’s tokens instead of chasing raw personal data across AI systems. Protecto also shows where sensitive data flows, which helps map where a request must reach. See GDPR compliance with Protecto and CCPA and CPRA compliance.

Frequently asked questions

How long do you have to respond to a DSAR?

Under the GDPR, one month, extendable by two more months for complex or numerous requests. Under the CCPA, 45 calendar days, extendable once by another 45 days with notice.

Can a company refuse a DSAR?

Under the GDPR, only if the request is manifestly unfounded or excessive, and the company must explain why. Under the CCPA, a business can deny a request it cannot verify.

Do AI chat logs count in a DSAR?

Yes, if they contain the person’s personal data. Prompts, chat histories and logs that relate to the person fall within the right of access, and so can other stored records derived from their data.

Related terms

Data subject · Right to be forgotten · GDPR · CCPA · Data mapping · Personal information

Book a demo

See Protecto on your own data

Book a demo to see how Protecto finds and masks sensitive data before it reaches any AI model.

New: Protecto now runs inside LiteLLM. Control what sensitive data your models and tools see.
Read the launch blog