CVE-2026-4870: IBM Qiskit SDK Parser Denial-of-Service Vulnerability

Technical summary

CVE-2026-4870 is a stack exhaustion vulnerability (CWE-674: Uncontrolled Recursion) in the IBM Qiskit SDK parser. The vulnerability exists in versions 0.43.0 through 2.5.0 and stems from insufficient recursion depth checks in the parser logic. When processing malformed or adversarially constructed input, the parser enters a recursive descent that exhausts the call stack, resulting in a segmentation fault. The vulnerability is remotely exploitable without requiring prior authentication or user interaction. The attack surface is present wherever the SDK's parser processes untrusted input.

Business impact

Organizations using the affected Qiskit SDK versions in production environments face availability risk. Quantum computing research teams, enterprises building quantum applications, and cloud services offering Qiskit-based tools could experience service interruptions if an attacker crafts and sends malicious input. This is particularly concerning for continuous research workflows or API services that automatically process user-supplied quantum circuits or configuration data. While confidentiality and integrity are unaffected, denial of service can disrupt research timelines, break dependent workflows, and damage service reputation.

Affected systems

IBM Qiskit Software Development Kit (SDK) versions 0.43.0 through 2.5.0 are vulnerable. This includes development environments, production deployments, cloud-hosted quantum computing services, and research platforms that bundle the affected SDK versions. Any system directly invoking the vulnerable parser with untrusted input is at risk. Downstream applications or services that depend on Qiskit are impacted only if they also include and expose the vulnerable SDK versions.

Exploitability

Exploitability is straightforward. The vulnerability requires only network access and no authentication, privilege escalation, or user interaction. An attacker with knowledge of the parser's input format can construct a payload that triggers deep recursion and craft a remote request (HTTP, API call, or direct connection) to deliver it. The low attack complexity and network accessibility result in a CVSS score of 7.5 (HIGH). However, practical exploitation depends on whether the vulnerable parser is exposed as part of an Internet-facing service or API.

Remediation

Organizations must upgrade to a patched version of the IBM Qiskit SDK. Verify against the official IBM Qiskit advisory for the specific version that resolves CVE-2026-4870; upgrades beyond 2.5.0 are expected to include a fix. If immediate patching is not possible, restrict network access to services running the SDK and validate or sanitize input before passing it to the parser. Consider implementing parser timeouts or recursion depth limits at the application level as a temporary mitigation.

Patch guidance

Update the IBM Qiskit SDK to a version newer than 2.5.0 that includes the security fix. Coordinate upgrades with development and operations teams to ensure compatibility with dependent code and quantum workflows. Test patched versions in a non-production environment first, particularly if custom quantum circuits or configurations are processed. Consult the official IBM Qiskit release notes and security advisories for patch availability and any breaking changes.

Detection guidance

Monitor for abnormal parser behavior or unexpected segmentation faults in Qiskit-based services. Implement input validation to reject or sanitize payloads that exhibit unusual nesting depth, recursion patterns, or malformed structure. Log and alert on crash events in Qiskit processes. If available, enable parser debugging or verbose logging to capture the characteristics of inputs that trigger crashes. Review network traffic for requests that deviate from normal quantum circuit submission patterns, especially those containing deeply nested or recursive structures.

Why prioritize this

This vulnerability should be prioritized if your organization runs Qiskit SDK in production, especially in user-facing or cloud-hosted services. The remote, unauthenticated exploitability and HIGH severity score necessitate timely patching. However, prioritization should be tempered by your specific deployment model: internal research environments with restricted network access face lower immediate risk than public APIs. Organizations with legacy Qiskit deployments in unsupported versions or without active maintenance should prioritize upgrade planning. This is not yet on the KEV (Known Exploited Vulnerabilities) list, so you have a window to patch before widespread exploitation.

Risk score, explained

The CVSS 3.1 score of 7.5 (HIGH) reflects a remotely exploitable denial-of-service condition with no authentication barrier. The attack vector (AV:N), low attack complexity (AC:L), and lack of privilege or user interaction requirements maximize accessibility. However, the impact is confined to availability (A:H); confidentiality and integrity are not compromised. The score does not account for deployed context: if your Qiskit service is not publicly exposed or processes only trusted input, actual risk is lower. Conversely, if Qiskit is part of a critical or high-availability service, organizational risk may exceed the base score.

Frequently asked questions