Quick answer
Every Data Fiduciary must keep personal data, associated traffic data and processing logs for at least one year from the date of processing under Rule 8(3) of the DPDP Rules, and keep security logs and personal data for one year under Rule 6(1)(e). This applies to processing by Data Processors too, survives account deletion, and starts on 13 May 2027. After a year, erase unless another law requires longer.
- Rule 8(3) sets a minimum one-year retention of personal data, associated traffic data and processing logs for Seventh Schedule purposes.
- Rule 6(1)(e) separately requires one year of logs and personal data for detecting, investigating and remediating unauthorised access.
- The obligation covers processing by Data Processors, so cloud, SaaS and AI vendors must keep logs for a year as well.
- Retention survives account deletion: the e-book illustration requires order and processing logs for a year even after the buyer deletes her account.
- After the year, data and logs must be erased unless another law or Government notification requires longer retention.
Who this is for: Platform, SRE and observability teams, Security operations, Data engineering teams, AI and LLM application teams, Cloud and SaaS vendors acting as Data Processors
What does DPDP require you to keep for one year?
Two separate provisions create a one-year retention duty, with overlapping but distinct scope.
| Rule 6(1)(e) | Rule 8(3) | |
|---|---|---|
| Context | Reasonable security safeguards | Retention and erasure |
| What | "Such logs and personal data" (the access logs in clause (c) and data needed for continuity) | Personal data, associated traffic data and other logs of the processing |
| Period | One year | Minimum one year from the date of processing |
| Purpose | Detecting unauthorised access, investigation, remediation, continued processing | Purposes in the Seventh Schedule |
| Exception | Unless another law requires otherwise | Longer if another law or Government notification requires |
| After the period | Not stated | Cause the data and logs to be erased |
Read together, from 13 May 2027 every Data Fiduciary must be able to produce, for any processing event in the past year, the personal data involved, the traffic data associated with it, and the logs showing what processing happened, including processing done by its Data Processors.
Rule 8(3) is expressly "without prejudice to sub-rules (1) and (2)", so it sits alongside the Third Schedule erasure rules for large platforms rather than replacing them. It was added in the final Rules; the January 2025 draft had no such floor.
Why: the Seventh Schedule purposes
Rule 8(3) ties the retention to "the purposes as specified in the Seventh Schedule". That schedule, also used by Rule 23 for Government information requests, lists three purposes:
- Use by the State or its instrumentalities of a Data Principal's personal data in the interest of sovereignty and integrity of India or security of the State. The authorised person is an officer of a notified instrumentality designated by the Central Government or the head of the instrumentality.
- Use by the State or its instrumentalities for performance of any function under law, or disclosure of information to fulfil any legal obligation. The authorised person is whoever is authorised under the applicable law.
- Assessment for notifying a Data Fiduciary or class as a Significant Data Fiduciary. The authorised person is a MeitY officer designated by the Secretary.
In other words, the one-year archive exists so that, if the Government lawfully calls for information under Rule 23, the data and logs still exist. That is why it applies even after a user deletes her account and why it extends to processors.
This purpose also limits what you may do with the archive. Data kept under Rule 8(3) is retained for Seventh Schedule purposes and security investigation, not for continued marketing, analytics or model training. See Rule 23 information requests.
Which logs to include: a practical inventory
The Rules do not list log types. A defensible inventory covers every layer where personal data is processed.
| Layer | Examples | Typical current retention |
|---|---|---|
| Application access logs | Who viewed or changed which customer record | Often 30–90 days |
| Authentication logs | Logins, MFA events, failed attempts | 90 days |
| API gateway and traffic logs | Request metadata, source IPs, user IDs, endpoints | 7–30 days |
| Transaction and event logs | Order, payment, delivery events (the e-book example) | Varies |
| Admin and privileged access logs | Production DB queries, support tool access | 90 days |
| Data pipeline logs | ETL jobs, exports, sharing with third parties | 30 days |
| AI and LLM logs | Prompts, retrieved context, tool calls, outputs tied to users | Often not retained or retained by vendor |
| Consent and rights logs | Consent given or withdrawn, rights requests and responses | Varies |
"Associated traffic data" is not defined in the Rules. Treat it as network and request metadata linked to processing: IPs, timestamps, session identifiers, device identifiers.
The common gap is application-level and AI logs. Infrastructure logs exist; logs of which person's data was processed, by whom, and why, often do not. See LLM logging under the one-year rule.
Making processors keep logs for a year
Rule 8(3) covers "any processing of personal data undertaken by it or on its behalf by a Data Processor". The cloud illustration makes the point directly: if company X hosts customer records with cloud provider C, X must ensure C also retains the data and associated logs for at least one year before erasure, unless another law requires longer.
That makes retention a procurement issue. For each processor:
- Check default retention. Many SaaS tools keep audit logs for 30 to 180 days on standard plans, with longer retention on premium tiers.
- Choose: vendor-side or export. Either contract for one-year retention at the vendor, or export logs to your own archive continuously. Export is often cheaper and keeps the evidence under your control.
- Update the contract. Add a clause requiring one-year retention of data and logs, erasure on instruction afterwards, and production on request. Rule 6(1)(f) already requires security provisions in processor contracts.
- Cover AI APIs. Model providers may retain prompts for limited periods or offer zero-retention modes. If you use zero retention at the vendor, you must keep the logs yourself.
- Test retrieval. Pick a random processing event from eleven months ago and confirm you can find its logs.
Do this before 13 May 2027 so the archive is already populated when the duty starts.
Protecting a year of logs
A year of logs is a year of personal data concentrated in one place. It is an attractive target and must meet Rule 6 itself.
- Minimise content. Log identifiers and actions, not full payloads. Record "user 123 viewed address record 456", not the address. Where full content is needed, as with AI prompts, tokenise or mask personal data before logging where you can.
- Encrypt the archive and hold keys separately, under Rule 6(1)(a).
- Restrict access to security, legal and compliance roles, with every archive query itself logged, under Rule 6(1)(b) and (c).
- Make it tamper-evident with write-once storage or hash chaining, so logs are credible evidence.
- Tier storage. Keep 30 to 90 days hot for operations and move the rest to cheaper cold storage. Retention does not require fast access.
- Synchronise clocks. CERT-In's 2022 Directions require clock synchronisation to NIC or NPL NTP servers or traceable sources; consistent timestamps matter for investigations.
CERT-In also requires ICT system logs to be kept securely for a rolling 180 days within Indian jurisdiction. The DPDP Rules do not add a localisation requirement for these logs, but if you are subject to CERT-In, keep at least the first 180 days in India and the full year under one policy.
Erasure requests versus the one-year floor
The most common question: what happens when a user asks for erasure, or withdraws consent, inside the one-year window?
The Rules answer through the illustration in Rule 8(3): the e-book platform must keep order details, personal data and processing logs for at least one year even if X deletes her account. Section 8(7) of the Act also makes erasure subject to retention necessary for compliance with law, and Rule 8(3) is such a requirement.
A practical approach:
- Act on the request immediately for live systems. Stop processing for the original purposes, remove data from product features, marketing, analytics and AI training.
- Move the relevant data and logs to restricted retention, accessible only for Seventh Schedule purposes and security investigation.
- Tell the user clearly that some records are kept for a legally required minimum period and will be erased afterwards.
- Erase on schedule once the year from the last processing passes, unless another law requires longer, and log that erasure.
After the year, Rule 8(3) says the fiduciary "shall cause such personal data and logs to be erased". Retaining logs indefinitely "just in case" is not compliant either. Set a firm upper bound per log type, informed by other laws such as tax and sectoral record-keeping.
Further reading on suhasbhairav.com
- BlogLLM Observability and Auditing in Production
- BlogAudit Logs for AI Agents: Traceable Actions in Production
- BlogProduction AI audit logs: prompt-response traceability
- AI LabBatuk sovereign AI chat workspace for Hash-chain audit evidence
- AI AutomationAI Automation for Security Monitoring: Uses + Example
Before 13 May 2027
Compliance checklist
Inventory every log source touching personal data
Include application, authentication, API, transaction, admin, pipeline, AI and consent logs.
Rule 8(3)
Extend retention to at least one year
Change log rotation and SaaS plan settings, or export to an archive, before 13 May 2027.
Rules 6(1)(e), 8(3)
Contract processors for one-year retention
Require processors to keep data and logs for a year and erase on instruction afterwards.
Rule 8(3), Case 2
Minimise log content
Log identifiers and actions rather than full personal data payloads; mask or tokenise where full content is needed.
Rule 6(1)(a)
Restrict and log archive access
Limit archive queries to security, legal and compliance and log every query.
Rule 6(1)(b), (c)
Implement restricted retention on account deletion
Remove data from live use but keep required records for one year.
Rule 8(3), Case 1
Set and enforce an upper bound
Erase data and logs after one year unless a specific law requires longer.
Rule 8(3)
Key dates
Recommended: start extending log retention so archives are populated by May 2027
Rules 6(1)(e) and 8(3) one-year retention duties apply
Answers
Frequently asked questions
How long must logs be kept under the DPDP Rules?
How long must logs be kept under the DPDP Rules?
At least one year. Rule 8(3) requires personal data, associated traffic data and processing logs to be retained for a minimum of one year from the date of processing, and Rule 6(1)(e) requires security logs and personal data to be kept for one year. Both apply from 13 May 2027.
Do I have to keep a user's data after they delete their account?
Do I have to keep a user's data after they delete their account?
Yes, for at least a year. The e-book illustration in Rule 8(3) says the platform must keep order details, personal data and processing logs for at least one year from the transaction even if the buyer deletes her account. Stop using the data for normal purposes and restrict access during that year.
Does the one-year log rule apply to my cloud provider?
Does the one-year log rule apply to my cloud provider?
Yes, through you. Rule 8(3) covers processing done on your behalf by a Data Processor, and its cloud illustration requires the fiduciary to ensure the cloud provider also retains data and logs for at least one year. Put this in the processor contract or export the logs yourself.
What is traffic data under DPDP Rule 8(3)?
What is traffic data under DPDP Rule 8(3)?
The Rules do not define associated traffic data. A sensible reading is network and request metadata linked to processing, such as IP addresses, timestamps, session and device identifiers and endpoints. Document your interpretation and apply it consistently.
Can I keep logs longer than one year under DPDP?
Can I keep logs longer than one year under DPDP?
Only if another law or Government notification requires it. Rule 8(3) says that after the minimum period the fiduciary shall cause the data and logs to be erased, unless further retention is required for compliance with another law. Indefinite retention is not compliant.
How does DPDP log retention compare with CERT-In?
How does DPDP log retention compare with CERT-In?
CERT-In's April 2022 Directions require ICT system logs to be kept securely for a rolling 180 days within Indian jurisdiction. The DPDP Rules require one year. Organisations subject to both should design to one year, keeping at least the CERT-In-scoped logs in India.
Legal references and sources
- Digital Personal Data Protection Rules, 2025 — Rule 8(3) with illustrations (official text)
- Digital Personal Data Protection Rules, 2025 — Rule 6(1)(e) (official text)
- Digital Personal Data Protection Rules, 2025 — Seventh Schedule (official text)
- CERT-In Directions under section 70B(6) IT Act — 28 April 2022 (official text)
This guide explains the law for engineering and business teams. It is not legal advice; confirm how it applies to your organisation with qualified counsel, and check the Gazette for later notifications or amendments.
About the author

Suhas Bhairav
Systems architect and applied AI practitioner who designs production AI systems for Indian and global companies, with a focus on privacy-by-design under the DPDP Act and the EU AI Act.
Related guides
- DPDP Rules 2025DPDP Data Retention and Erasure Under Rule 8 and the Third ScheduleRule 8 sets when personal data must be erased: three-year dormancy erasure for large e-commerce, gaming and social media platforms, a 48-hour warning, and a one-year retention floor for everyone.
- AI & personal dataLLM Logging and Observability Under DPDP's One-Year RuleDPDP asks you to minimise data yet keep processing logs for a year. A tiered design for LLM traces resolves the tension: log less, mask early, lock access, then erase on schedule.
- DPDP Rules 2025DPDP Reasonable Security Safeguards Under Rule 6Rule 6 lists the minimum security safeguards every Data Fiduciary must take from 13 May 2027. Failure can attract a penalty of up to ₹250 crore, the highest in the DPDP Act.
- India AI & sector rulesCERT-In Rules for AI Systems: Incident Reporting & LogsHow CERT-In's April 2022 directions on 6-hour incident reporting, 180-day logs in India and clock sync apply to LLM gateways, model APIs and AI agents, and how they meet DPDP.
- DPDP Rules 2025DPDP Rule 23: Government Calls for Information and the Seventh ScheduleRule 23 lets the Central Government require Data Fiduciaries and intermediaries to furnish information for three Seventh Schedule purposes, sometimes with a gag on telling users.
- DPDP Rules 2025Algorithmic Due Diligence for Significant Data FiduciariesRule 13(3) requires Significant Data Fiduciaries to verify that technical measures, including algorithmic software, are not likely to pose a risk to Data Principals' rights. Here is what that means in practice.