Fortinet FortiSandbox EUVD-2026-35443

The attacker needs only network reachability to the FortiSandbox HTTP service and the ability to send a specifically crafted HTTP request - no credentials, no user interaction, and no non-default toggle is called out in the advisory, consistent with PR:N/UI:N/AC:L in the CVSS vector. The vulnerable software must be FortiSandbox on-premise 5.0.0-5.0.5, 4.4.0-4.4.8, or any 4.2.x build, or FortiSandbox Cloud/PaaS 5.0.4-5.0.5. The practical limiting factor is exposure of the management/API HTTP interface: deployments that restrict that interface to an isolated management VLAN or jump host substantially reduce who can reach the vulnerable endpoint, whereas any FortiSandbox whose web interface is reachable from broader corporate networks or the internet is directly exploitable.

The CVSS 3.1 vector AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H describes the worst-case profile: remotely reachable, low complexity, no authentication, no user interaction, and full confidentiality/integrity/availability impact - consistent with a 9.8 critical score and with the description's wording of an unauthenticated attacker reaching the HTTP endpoint. No EPSS score, KEV listing, or SSVC decision data was supplied, and no public proof-of-concept is identified at time of analysis, so the real-world exploitation tempo is unknown; however, Fortinet appliances have a strong historical track record of post-disclosure mass exploitation, and a pre-auth RCE on a security appliance is exactly the class of bug that ransomware and state-aligned operators have weaponized within days of patch release in past Fortinet advisories. Two caveats worth flagging: the CVSS vector implies the vulnerable HTTP service must be network-reachable from the attacker, which in well-segmented deployments may be limited to management VLANs, and the lack of any third-party reference beyond Fortinet's own PSIRT page means independent technical detail (root-cause endpoint, payload shape, indicators) was not available for this synthesis.

An attacker who can reach the FortiSandbox HTTP interface - for example a compromised internal host, or an internet-exposed management portal - sends a crafted HTTP request whose parameters contain shell metacharacters that the vulnerable handler passes unsanitized to an OS command, yielding code execution under the service account. From that foothold the attacker can read submitted sample files and analysis verdicts, tamper with detection logic to whitelist their own malware, and pivot into the management fabric the sandbox is trusted to talk to (FortiGate, FortiAnalyzer, mail gateways). No public proof-of-concept is identified at time of analysis, but the low attack complexity in the CVSS vector implies reliable single-request exploitation once the vulnerable endpoint is identified.