priyaneelkrish
ServiceNow Employee

priyaneelkrish_0-1795200625523.png

Time-bound entity exceptions in Access Intelligence

 

Our past posts followed a finding on its way to a fix: remediation that fits the finding, a blast radius preview before you revoke, and Query Agent to explain a query in plain language. All three assume the risk should go away. Sometimes it shouldn’t. Not yet, anyway.

 

Every security team grants exceptions. The break-glass admin account that has to keep standing access. The contractor who needs delete rights until the data migration wraps. The service account that can’t enroll in MFA until a vendor ships a fix. Accepting a risk for a while is a legitimate call, and it’s how real organizations keep working.

 

The trouble starts after the call is made. The approval lives in a ticket comment or a chat thread. The reason lives in one person’s head. And because nobody set an end date, the exception never ends. Months later the person who granted it has changed teams, the migration finished in the spring, and the account still sits outside the rules with no one watching.

 

An exception without an end date isn’t an exception. It’s a policy change nobody approved.

 

Entity exceptions in Access Intelligence now carry a reason, an expiry date, and the name of the person who granted them. You grant them where the risk lives, change many at once without guesswork, see every change in a history, and watch exactly how they shape an identity’s risk score. When the date arrives, the exception ends and the risk counts again.

 

The exception everyone forgot

Some of the most damaging breaches of the past few years didn’t need a clever exploit. They needed one account, one tenant, or one credential that someone had decided was fine to leave alone.

 

The VPN profile nobody was using

In May 2021, ransomware forced Colonial Pipeline to shut down its 5,500-mile fuel pipeline. Its CEO later told the Senate that the attackers got in through a legacy VPN profile that was not intended to be in use, protected by a single password and no second factor. The account had outlived its purpose. Nobody had decided to keep it. It simply never had a date to go away.

 

The test tenant that lived outside the rules

In January 2024, Microsoft disclosed that the threat actor Midnight Blizzard had used a password spray to compromise a legacy, non-production test tenant account that didn’t have MFA. From there the actor found a legacy test OAuth application with elevated access to Microsoft’s corporate environment. Microsoft noted that if the same team deployed that tenant today, mandatory policy would ensure MFA was on. Test environments are the classic quiet exception. It’s just test, until it isn’t.

 

Four credentials believed to be unused

After a breach at a third-party identity provider in October 2023, Cloudflare set out to rotate every credential that had been exposed. It rotated thousands. It missed one service token and three service accounts because it was mistakenly believed they were unused. In mid-November, a threat actor used that token and one of those accounts to reach Cloudflare’s Atlassian environment. Strong access controls limited how far the attacker got, and the team spotted the intrusion on Thanksgiving Day. Still, “unused” had quietly become an exception with no owner, no review, and no end date.

 

Three organizations with serious security teams. Three industries. One pattern. Someone decided an account, a tenant, or a credential was fine to leave alone, and that decision had no reason on record and no date to revisit it. More detection wouldn’t have changed that. An end date would have forced the question.

 

Accepted risk piles up quietly

This isn’t a handful of unlucky companies. Known problems stay open, and they stay open for a long time.

The Verizon 2026 Data Breach Investigations Report found that only 23% of third-party organizations fully fixed missing or weak MFA on their cloud accounts. For weak passwords and permission misconfigurations, it took almost eight months to resolve just half of the findings. Even the most urgent work slipped: organizations fully remediated only 26% of the critical vulnerabilities on CISA’s Known Exploited Vulnerabilities catalog in 2025, down from 38% the year before, and the median time to a full fix grew from 32 to 43 days.

 

priyaneelkrish_1-1795200691217.png

Fixes take longer than most plans assume, which is exactly why paused risks need a return date.

 

The cost of a miss keeps climbing too. IBM’s 2026 Cost of a Data Breach Report puts the global average breach at a record $4.99 million, and $11.5 million in the United States.

 

None of this means every open finding is negligence. Backlogs are real, and some risks have to wait. What matters is whether the risks you’ve chosen to live with are tracked as decisions or forgotten as defaults.

 

Standards already expect temporary access to end on its own. NIST SP 800-53 control AC-2(2) calls for temporary and emergency accounts to be automatically removed or disabled after a period the organization defines. A risk exception is the same kind of promise. It deserves the same kind of timer.

 

Where exceptions go wrong

Every exception passes through the same moments. Each one is a place where a reasonable decision can quietly turn into permanent risk.

Where it breaks

What usually happens

What you get now

Granting it

Approval happens in chat or a ticket, far from the finding

Mark as exception right on the risk, from the query, the graph, or the risk score

Remembering why

The reason lives in one person’s head and leaves when they do

A reason saved with the exception and shown wherever the risk appears

Ending it

No end date, so it never ends

30, 60, or 90 days, or any date you choose. No expiry is still an option, but now it’s a choice

Changing many

One at a time, or a bulk edit that overwrites reasons you meant to keep

Bulk mark and edit with a live summary of exactly what will change

Proving it

An auditor asks who approved it and when, and nobody can say

Exception history for every entity: granted, updated, removed, with old and new values

Scoring it

Excepted risks keep inflating scores, or disappear without a trace

Paused risks stay visible in the risk score panel, labeled with the date they return

 

What’s new: entity exceptions with an expiry date

Exceptions now work at the entity level across Access Intelligence. Each one records why it was granted, when it ends, and who granted it. You can manage them from a query’s Affected Entities tab, from an identity in Graph Builder, or straight from an identity’s risk score. Wherever you start, it’s the same exception with the same history.

 

Here’s the flow end to end, following one Okta identity, Douglas Holland, through a Critical query.

 

Start with the risks you’ve already accepted

The Risks page now shows an Exceptions count beside Affected Entities for every query. Identities with MFA Disabled, for example, flags 13,645 entities and carries 10 exceptions. You can see which findings have accepted risk inside them before you open a single query.

 

priyaneelkrish_2-1795200747620.png

Exceptions sit next to affected entities, so accepted risk never hides inside a total.

 

 

See who’s excepted, why, and until when

Open Okta users with delete access permissions, a Critical query under the Privileged Access risk profile. The Affected Entities tab lists 13,401 risks and exceptions, with four columns that tell the story: Exception, Exception Expires, Excepted By, and Exception Reason. A filter lets you show affected entities, exceptions, or both.

 

The first three rows say a lot. One exception expires in 3 months with the reason bulk omit. Another has a reason, ignore, but no end date. The third has neither. Those last two are exactly the exceptions this feature exists to catch.

 

priyaneelkrish_3-1795200783582.png

Expiry, owner, and reason, right beside every excepted entity.

 

 

Mark an exception with a reason and an end date

Choose Mark as exception and add a reason. Then set when it ends: 30, 60, or 90 days, a specific date on the calendar, or no expiry. Past dates are grayed out, so you can’t create an exception that has already expired.

 

We marked Douglas Holland with the reason ignore for now and an end date of today. The table updated in place. Exception Expires now reads in 8 hours, because the exception runs through 11:59 PM on the date you pick, and the reason and the name of the person who granted it sit right beside it. Everything else is one click away in the row menu: edit the exception, unmark it, view its history, or open the entity in Graph Builder or Query Builder.

 

Change many at once, and know exactly what will happen

Bulk changes are where exception hygiene usually falls apart. Select a mix of entities and the dialog tells you what you’re working with, in this case 3 selected, 2 new, and 1 already an exception. Then it spells out the result before you commit: 2 risks become exceptions expiring February 18, 2027, and 1 existing exception has its reason removed and takes the same expiry.

 

Two checkboxes keep you from overwriting what you meant to keep. Set or edit reason gives every selected risk the same reason, and leaving it empty removes it. Set or edit expiry gives every selected risk the same end date. Leave either one unchecked and existing exceptions keep what they already have. No guessing, and no cleanup afterward.

priyaneelkrish_5-1795200878153.png

The summary updates as you choose, so a bulk edit never surprises you.

 

Every change leaves a trail

Choose View Exception History on any entity to see the full story. Douglas Holland has been excepted 3 times. Each entry shows who acted, when, and whether the exception was granted, updated, or removed. Expand an update to compare old and new values: here the reason changed from we know this is ok to ignore, and the expiry moved up from February 18, 2027 to December 20, 2026. When an exception is removed, the entry says it plainly. The risk is active again.

 

That’s the auditor’s favorite question, answered before anyone asks it.

priyaneelkrish_6-1795200917542.png

Granted, updated, removed. Every change, with the before and after.

 

Manage exceptions from the identity, not just the query

Queries are one way in. Identities are the other. Open Douglas Holland in Graph Builder and the side panel lists every risk he matches, each with its severity. Excepted risks carry an Exception badge, with a clock icon when the exception has an end date. Mark, edit, or unmark right there.

 

See how exceptions shape the risk score

Select the risk score to open the Risk Score panel, which explains the number in plain language. Douglas Holland has a Critical score of 78 out of 100, the cumulative result of 9 active risk signals. Excepted risks don’t vanish from view. They’re grayed out and labeled, for example Omitted until 2027-01-19 at 23:59:59, so you always know which risks are paused and when they come back.

 

Bulk Edit works here too. We selected four risks, one new and three already excepted, and set them all to end on December 20, 2026. The panel confirmed it with 1 query omitted and 3 omissions updated. The count dropped to 8 active risk signals, and a note flags that the score above may not reflect the latest changes yet.

 

Time-bound entity exceptions Demo 

 

How this strengthens your identity security posture

Accepted risk becomes a decision, not a default. Every exception carries a reason, an end date, and the name of the person who granted it, recorded right where the risk lives.

Exceptions end on schedule. Quick picks for 30, 60, and 90 days make a time limit the easy path. No expiry is still there when you need it, but now someone has to choose it.

Bulk changes don’t backfire. The live summary shows what happens to new and existing exceptions before you confirm anything.

Audits get easier. Exception history answers who, when, why, and what changed for every entity.

Risk scores stay honest. Paused risks stay visible with their return date, so a lower score never hides a problem you’ve only postponed.

You can start anywhere. Query, graph, or risk score, it’s the same exception and the same history.

 

Where this is going

Exceptions with an end date reflect a simple idea we’re building into Access Intelligence: people make the judgment calls, and the platform remembers them. Accepting a risk is a decision that belongs to your team. Keeping track of when that decision runs out shouldn’t depend on anyone’s memory.

 

We’re extending the same pattern across Access Intelligence, so the path from accepted risk back to an active fix stays in one place, with your team in the loop.

 

Try it today

Open any query in Access Visibility, go to Affected Entities, and set the filter to show exceptions. Look for rows where Exception Expires reads Not Set. Ask whether each one still needs to be there. Give the ones that do a reason and an end date, and let the rest go.

 

Every exception is a promise to come back. Now that promise has a date on it.

 

About the author

priyaneelkrish_8-1795201310110.png  Priyanka Neelakrishnan is a Senior Principal Product Manager at ServiceNow, where she leads product strategy and execution for Access Security on the Veza Access Graph, spanning identity security, closed-loop remediation, identity risk and adaptive role studio. She previously built enterprise security and data platforms at Palo Alto Networks and Symantec and holds a US patent in data and access security, with another filed. She is the author of multiple books on data and AI security, including Autonomous Data Security and Jailbreaking LLMs, and writes and speaks widely on modern enterprise security.

 

For more information

www.servicenow.com