Safeguard
Compliance•

Do you have to tell customers if a bad package never ran in production?

Not automatically. Notification usually depends on whether personal data or customer systems were affected. How to decide, and what to document anyway.

Safeguard Security Team
3 min read

Not automatically. Whether you must notify customers depends on your contracts and on the law that applies to you, which usually turns on whether customer data or systems were affected. If a bad package never ran in production and nothing reached customer data, notification is often not triggered, but you should document why. This is not legal advice.

What the law actually keys on

Take GDPR as one example. Article 33 requires a controller to notify the supervisory authority without undue delay and, where feasible, within 72 hours of becoming aware of a personal data breach, unless the breach is unlikely to result in a risk to people's rights and freedoms. Even then, the controller must document the breach. A processor must tell the controller without undue delay. The trigger is a personal data breach, not the presence of a malicious package.

A decision path

  1. Did the bad version ever execute? If not, stop here and document.
  2. Where did it execute: developer machine, CI, staging or production?
  3. What could that environment read: customer data, production credentials, secrets that unlock customer data?
  4. Is there evidence of exfiltration or misuse?
  5. Do your customer contracts set their own notice duties, which are often stricter than the law?
SituationTypical outcome
Present in a lockfile, never builtNo notification, record the search
Ran in CI with no production accessRotate secrets, internal record, often no notice
Ran in CI with production credentialsAssume exposure, assess data access, may notify
Ran in production or exfiltration seenLikely notification duties, involve counsel

Document your reasoning

Write a short record: what you found, the window, the evidence, the decision and who approved it. A regulator or customer reviewing later cares less about the answer than whether you reasoned from evidence. See how to answer when a customer asks if you were affected.

Voluntary notice

Some teams tell customers even when not required, to show control. If you do, keep it factual and short, and agree it with counsel first.

Contracts often decide first

Many customer agreements set notice periods and triggers that are shorter or broader than the law. Read the clauses before you decide, and check whether they refer to security incidents in general or only to confirmed data breaches. If you act as a processor, you also owe your customers an alert without undue delay once you become aware of a breach. Sales and legal should know about the clause list before an incident, not during one.

Quick check

Does a malicious dependency count as a personal data breach?

Only if it led to unauthorised access, loss or disclosure of personal data. Presence alone does not.

Who decides whether to notify?

Usually legal and security leadership together, using the evidence record and contract terms.

Safeguard's compliance and evidence automation can help keep records of controls and decisions organised. You can try a free scan at gold.safeguard.sh/scan.

Sources

Last reviewed: 19 October 2026.

Never miss an update

Weekly insights on software supply chain security, delivered to your inbox.

Self-healing security runs on Safeguard.

Your first fix PR is minutes away.

No sales call required, even your agent can complete the purchase over MCP.