Coordinated vulnerability disclosure

Also known as: CVD

On this page 7

Coordinated vulnerability disclosure (CVD) is the structured process in which the person who finds a security flaw, the vendor responsible for the affected product and, where needed, a neutral coordinator work together to fix the flaw before its details become public. In cybersecurity, CVD is the dominant norm for handling newly discovered vulnerabilities, because it balances two goals that pull in opposite directions: giving defenders time to patch and giving the public accurate information about risk.

The model replaced the older term “responsible disclosure” in most policy documents, since the newer name stresses shared obligations among all parties rather than a duty placed on the researcher alone. It is closely tied to other concepts in the Vulnerabilities topic, such as CVE identifiers, CVSS severity scores and vulnerability management programs.

How coordinated vulnerability disclosure works

A CVD process moves a vulnerability report through a sequence of defined phases, from first contact to public advisory. Each phase has an owner and an expected duration, and the clock usually starts when the vendor acknowledges the report, not when the flaw was found.

The table below summarizes the typical phases and what each party does in them.

PhaseMain actorKey activity
DiscoveryFinderIdentifies the flaw and documents reproduction steps
ReportingFinderSends a private report through the vendor’s published channel
Validation and triageVendorReproduces the issue, assesses impact and assigns severity
RemediationVendorDevelops, tests and packages a fix or mitigation
Coordinated releaseVendor and coordinatorPublishes patch, advisory and CVE record on an agreed date
Post-release follow-upAll partiesAnswers questions, credits the finder and reviews the process
Typical phases of a coordinated disclosure

During remediation, the parties exchange information under embargo, meaning that nobody discloses technical details until the agreed date. When several vendors share a vulnerable component, a multi-party coordination effort aligns all their release schedules so that no single advisory exposes the others’ customers.

Roles and supporting standards

CVD defines four recurring roles: the finder, the vendor, the coordinator and the deployer who must install the fix. Clarity about roles prevents the most common breakdown, which is a report that reaches nobody able to act on it.

RoleTypical examplesMain responsibility
Finder or reporterIndependent researcher, academic team, customerReport privately and avoid harm while testing
VendorSoftware maker, hardware manufacturer, open-source maintainerAcknowledge, fix and publish an advisory
CoordinatorNational CSIRT, CERT/CC, bug bounty platformMediate, locate contacts and enforce deadlines
DeployerIT operations team, cloud customer, device ownerApply patches or mitigations after release
Participants in the disclosure process

Several published standards formalize the process. ISO/IEC 29147 covers how vendors receive reports and publish advisories, while ISO/IEC 30111 covers internal handling and remediation. The security.txt file defined in RFC 9116 gives researchers a machine-readable contact address, and a vulnerability disclosure policy (VDP) states the scope, rules and safe-harbor commitments of an organization.

Regulation increasingly makes CVD mandatory rather than voluntary. The European Union’s Cyber Resilience Act requires manufacturers of products with digital elements to maintain a disclosure policy, and the NIS2 Directive assigns national CSIRTs a coordinator role for cross-border cases. United States federal agencies must publish a VDP under a binding operational directive from CISA.

How CVD compares with other disclosure models

Disclosure models differ mainly in who controls the timing and whether the public learns of the flaw before a fix exists. CVD sits between the extremes of full secrecy and immediate publication.

ModelTiming of public detailsMain risk
Coordinated disclosureAfter a fix or an agreed deadlineVendor may stall within the deadline
Full disclosureImmediately, without vendor noticeAttackers gain a head start over defenders
Private or nondisclosureNever, or only to the vendorUsers never learn they were exposed
Bug bounty programPer program rules, often coordinatedConfidentiality terms can silence researchers
Vulnerability broker saleUsually neverFlaw becomes an exploited zero-day
Disclosure models compared

A bug bounty is best understood as a CVD channel with financial rewards, not as a separate philosophy. Full disclosure remains a legitimate fallback when a vendor ignores reports for months, and many researchers treat it as the credible threat that keeps vendors engaged.

Disclosure deadlines and timing decisions

Most coordinators publish a default disclosure deadline, after which details become public whether or not a patch is ready. The best-known figure is 90 days, used by several industry research teams, while CERT/CC has long used 45 days as its baseline; both allow extensions when a vendor shows real progress.

The following decision table shows how practitioners usually adjust the timeline to circumstances.

SituationTypical adjustmentReason
Active exploitation observedShorten to daysUsers need mitigations immediately
Vendor unresponsive after repeated contactEscalate to a coordinator, then publishSilence cannot block public awareness
Complex hardware or firmware fixExtend by agreementPatch cycles and recertification take longer
Many vendors share the componentExtend and synchronizeOne early advisory would expose others
Fix released before the deadlinePublish with the fixDetails help defenders prioritize patching
Adjusting the disclosure timeline

Timing also interacts with patch adoption. A common practice is a short grace period of a few weeks after the patch ships before releasing proof-of-concept code, so that deployers have time to update.

Common failure modes in disclosure programs

Coordinated disclosure fails most often through organizational gaps rather than technical ones. Experienced coordinators repeatedly see the same patterns:

  • No published contact channel, so reports go to sales or general support addresses and disappear.
  • Legal threats against researchers, which discourage future reports and push finders toward full disclosure or brokers.
  • Silent patching, in which a fix ships without an advisory or CVE, leaving deployers unable to judge urgency.
  • Leaks during embargo, often through public code commits or an overly wide distribution list.
  • Severity downplaying, where the vendor assigns a low CVSS score that does not reflect realistic attack scenarios.

Mature programs counter these problems with a clear safe-harbor statement, a dedicated product security incident response team (PSIRT), defined service levels for acknowledgment, and honest advisories that credit the finder.

The simple explanation

Imagine noticing that the lock on a shop’s back door is broken. You could shout about it in the street, which might invite burglars, or you could quietly tell the owner, give them time to replace the lock, and then let neighbors know so they can check their own doors. Coordinated vulnerability disclosure is the second approach, applied to software and devices.

The person who finds the flaw tells the company privately, the company fixes it, and then everyone announces the problem and the fix together. The key idea is a fair deadline: the company gets time to repair the weakness, but not unlimited time to hide it.

How it is used in practice

A university researcher discovers that a widely used network router allows remote code execution through a crafted web request. She finds no security.txt file and no reply from the manufacturer’s support address after two weeks, so she sends the report to her national CSIRT, which locates the manufacturer’s product security team and opens a coordinated case with a 90-day deadline.

The manufacturer reproduces the flaw within a week, rates it critical, and learns that the vulnerable code comes from a third-party web server library also used by two other router makers. The coordinator brings all three vendors into one embargoed case and extends the deadline by 30 days to synchronize firmware releases.

On the agreed date, all three vendors publish firmware updates and advisories, a CVE record goes live, and the researcher is credited. She waits three more weeks before presenting technical details at a conference. The outcome is a fix for every affected customer before attackers learned the details.