How do you use Gmail content compliance rules to improve email security in Google Workspace?

Gmail's built-in filtering catches most spam and malware without any configuration. Compliance rules are how you add your own judgment on top: quarantine the patterns that hurt your business, tag what deserves a second look, and reject what you never want. They live under one page in the Admin console and share a common rule format, which makes them flexible and easy to get wrong.

This guide covers the three rule types, how Gmail evaluates them together, and three security rules worth copying. For Gmail's own phishing and spoofing protections, see our advanced phishing and malware protection guide first. Those settings should be on before you write custom rules.

Where are the compliance settings?

In the Google Admin console, go to Menu > Apps > Google Workspace > Gmail > Compliance. You need the Gmail Settings administrator privilege (Google Workspace Admin Help). Select an organizational unit on the left if the rule shouldn't apply to everyone, then hover over a setting and select Configure, or Edit or Add another if rules already exist.

Three settings on that page do the security work:

Each rule gets a unique description, a choice of messages to affect, up to 10 expressions, and one action.

How do content compliance expressions work?

Messages to affect

You can apply a rule to inbound, outbound or internal messages. Google counts a domain as internal if it's a verified Workspace domain, or a subdomain or parent domain of one (Google Workspace Admin Help). For security rules on incoming mail, tick Inbound only, so you don't trip on your own staff's messages.

Match types

Choose whether any or all of the expressions must match (Google's page shows the "If ANY of the following match the message" option), then add expressions:

• Simple content match works like Gmail search, so loose phrases match more than you'd expect.
• Advanced content match picks a location and a match type. Locations include Headers and body, Full headers, Body, Subject, Sender header, Recipients header, Envelope sender, Any envelope recipient and Raw message. Match types include Contains text, Equals, Matches regex, Not matches regex, Matches any word and Matches all words.
• Metadata match checks properties rather than text: Message authentication, Source IP, Secure transport (TLS), S/MIME status, Message size, Gmail confidential mode, and, in editions with Security Sandbox, malware detected by the sandbox.
• Predefined content match uses Google's detectors, such as US credit card or Social Security numbers, with an optional count and confidence threshold. It isn't available in every edition.

Google defines "authenticated" for metadata matching in DMARC terms: the message passes SPF with the envelope sender aligned to the From domain, or passes DKIM for the From domain. That's a useful building block, because it lets a rule react to spoofing.

Regex tips

Regex uses Google's RE2 syntax, and each expression is limited to 10,000 characters. Two details from Google's documentation catch people out:

• Full headers are checked one header field at a time, so one regex can't span the To: and Cc: fields.
• For content filtering, Gmail removes dots and plus signs from usernames, so jane.doe@ in a rule may never match.

Add (?i) at the start of a pattern to make it case-insensitive. In RE2, . doesn't match a line break unless you also add the s flag, as in (?is), which matters when a phrase wraps across lines.

What gets scanned

Content compliance scans text attachments and common types such as .docx, .xlsx and .pdf, and "any rule that applies to the message body text also applies to the extracted text" (Google Workspace Admin Help). Google's attachment compliance page sets the limits: content filters scan up to 1 MB of converted content, Gmail "doesn't convert or scan files that are larger than 10MB," and it can't open password-protected files or archives. Google also warns that "there might be false positives and negatives." Treat rules as a useful layer, not a guarantee.

Which actions can a rule take?

Avoid one modify option: Bypass spam filter for this message. It delivers incoming mail even when Gmail says it's spam. Attackers love any rule that grants a bypass based on text they can write.

For quarantine actions, create a custom quarantine first so each rule's catches land somewhere meaningful. Our Gmail admin quarantine guide covers setting them up and reviewing them.

In what order does Gmail evaluate rules?

Gmail doesn't run rules one after another and stop at the first match. Google describes this process (Google Workspace Admin Help):

1. Gmail "evaluates the message against all policies independently" and builds a list of consequences.
2. "If there are any reject consequences, Gmail rejects the message and ignores the other consequences."
3. Otherwise, Gmail checks for conflicts. These only arise when two consequences would change the primary route or the primary envelope recipient.
4. Conflicts are settled by specificity ("locally applied settings take precedence over inherited ones"), then by age ("older settings take precedence over newer ones").
5. All non-conflicting consequences are applied together. Two different subject prefixes are both added.

Google adds that "there's no strictly defined ordering of conflicting rules between different sections." When a message matches several quarantine rules, the quarantine documentation says it's sent to all the quarantines those rules name.

The practical upshot: you can't build an "allow first, then block" chain. To exempt a sender, use the rule's own Show options > address list bypass, which Google says bypasses the setting entirely on a match. Keep Authentication required on for those entries, or a forged From address gets the exemption too.

Three security rules worth adding

These three rules are our own examples, not Google templates. Treat each pattern as a starting point to test on a small organizational unit, as described below, before you quarantine real mail.

1. Quarantine invoices that announce changed bank details

Payment redirection is one of the costliest email frauds, and the messages often contain no link or attachment for filters to catch. See our guide to vendor invoice fraud.

1. Create a custom quarantine named Payment change review, with your finance lead's group as reviewers.
2. Add a Content compliance rule. Description: Inbound: changed bank details. Messages to affect: Inbound.
3. Set the rule so that all expressions must match and add two expressions:
   • Advanced content match, location Body, Matches regex: (?is)\b(new|updated|changed|different)\b.{0,40}\b(bank|banking|account|remittance|payment)\b.{0,20}\b(details|information|account)\b
   • Advanced content match, location Headers and body, Matches any word: invoice remittance payment wire IBAN
4. Action: Quarantine message, pointing at Payment change review.
5. Under Show options, add an address list of the few suppliers who legitimately send these notices, set to bypass, with authentication required.

Because body rules also apply to text extracted from attachments, this catches the wording inside a PDF invoice too. Expect some legitimate matches. That's the point: a finance person reviews them and confirms changes by phone, using a number already on file.

2. Quarantine risky attachment types that Gmail still allows

Gmail already blocks executables and scripts such as .exe, .js, .vbs, .ps1, .iso and .lnk, including inside archives (Gmail Help), and Google says file types "detected to be executables are automatically rejected." Your rule should target the next tier: formats attackers use to carry links and smuggle payloads past those blocks. Our guides to dangerous attachment types and HTML smuggling explain why.

1. Add an Attachment compliance rule. Messages to affect: Inbound.
2. Choose If ANY of the following match the message, add an expression of type File type, and enter custom file types such as html, htm, shtml, svg. Tick Also match files based on file format so renamed files are caught too. Google matches extensions it doesn't list as supported, such as .htm, by file name only.
3. Action: Quarantine message to a quarantine named Risky attachments.

For the file types Google supports, Gmail can detect the type from the file's content, so renaming the extension doesn't help an attacker. Gmail also checks the names of files inside archives. It can't look inside password-protected archives, so consider a second expression that quarantines encrypted attachments from outside, if your business rarely receives them.

3. Tag or quarantine unauthenticated mail claiming to be from your domain

1. Add a Content compliance rule. Messages to affect: Inbound.
2. Set the rule so that all expressions must match and add:
   • Advanced content match, location Sender header, Matches regex: (?i)@(example\.com|.+\.example\.com)$ (use your own domain)
   • Metadata match, attribute Message authentication, Message is not authenticated
3. Action: Quarantine message, or Modify message with Prepend custom subject set to Unverified sender while you test.

Gmail's Safety page already offers Protect against inbound emails spoofing your domain, so turn that on first. This rule gives you your own quarantine and wording for the same pattern, and the regex covers your subdomains too. Make sure your own SPF, DKIM and DMARC are right before you quarantine, or your own third-party senders will land there. Check them with the DMARC Checker.

A note on tagging all external mail

A content compliance rule can prepend a subject tag such as [External] to every inbound message. It works, but subject tags clutter threads and users stop noticing them. Gmail already shows warning banners for many risky situations, which our Gmail warning banners guide explains. Use a custom header instead if another system needs to know the message came from outside.

How should you test a new rule?

1. Create the rule on a test organizational unit with one or two accounts.
2. Use Modify message with a custom header, such as X-Rule-Test: payment-change, instead of quarantine at first.
3. Send test messages from an outside account, then check the headers with the Email Header Analyzer or trace them with Email Log Search.
4. Allow up to 24 hours for changes to apply everywhere.
5. Switch to quarantine or reject only when a week of real traffic gives acceptable matches.

What compliance rules can't do

Rules match patterns. They don't judge reputation, detonate links or learn from user reports, and they only see mail after it reaches Gmail. If your MX points to a filtering gateway in front of Google Workspace, attachment and content policies can run there first, and Gmail's rules add a second layer. Our inbound gateway guide covers that setup, and our Google Workspace platform page explains how iBlockSpam, with its attachment-type filters and custom domain rules, sits in front of Gmail.

Next steps

• Turn on the Safety settings in Gmail first, then add the bank-details and attachment rules above to a test organizational unit.
• Create custom quarantines so each rule's catches are easy to review.
• Review all existing compliance rules for any that use Bypass spam filter for this message or address lists without authentication.
• Revisit your Gmail spam settings so approved senders and compliance exemptions don't overlap.

See all spam filter setup guides.