CVE-2026-14120
Inappropriate implementation in DevTools in Google Chrome prior to 150.0.7871.47 allowed a remote attacker who had compromised the renderer
CVSS
9.6
Critical
EPSS
0.3%
p28
KEV
—
Exploit Today
8
0-100
Published: Jun 30, 2026 · Last modified: Jul 2, 2026 · CWE-20 · CWE-693
0.3%EPSS · 30 days0.3%
2026-08-182026-09-15
Inappropriate implementation in DevTools in Google Chrome prior to 150.0.7871.47 allowed a remote attacker who had compromised the renderer process to potentially perform a sandbox escape via a crafted HTML page. (Chromium security severity: Low)
CVECVSSEPSSKEVRExploitTitleMod.
CVE-2026-918424.1 MED—
——0A vulnerability has been found in OpenBankProject OBP-API up to 1.10.1. This impacts the function KryoInjection.invert of the file obp-api/src/main/scala/code/api/cache/Redis.scala of the component Kryo Handler. Such manipulation leads to deserialization. The attack can be launched remotely. A high complexity level is associated with this attack. The exploitability is said to be difficult. The exploit has been disclosed to the public and may be used. The project was informed of the problem early through an issue report but has not responded yet.22hCVE-2026-54254——
——0Cyberdrop-DL is a bulk asynchronous downloader for multiple file hosts. From 8.5.0 until 9.14.0, the Pixeldrain crawler uses substring host matching instead of requiring the input host to be an exact member of SUPPORTED_DOMAINS, and then reuses that input host for API requests. When a Pixeldrain API key is configured, processing a crafted URL from an attacker-controlled lookalike host causes cyberdrop_dl/crawlers/pixeldrain.py to send the Authorization header to that host. The URL may be introduced through a third-party site that can cause downloads for other sites, and the attacker receives the user's Pixeldrain API key. This issue is fixed in version 9.14.0.23hCVE-2026-47780——
——0free5GC is an open-source implementation of the 5G core network. In 4.2.3 and earlier, HandleCreateEeSubscriptions and HandleQueryeesubscriptions in free5gc/udr internal/sbi/api_datarepository.go validate the ueId path value with a regular expression whose final .+ alternative accepts every non-empty string instead of restricting identifiers to supported SUPI and GPSI formats. An actor with network reachability to the UDR SBI can submit a non-3GPP identifier to /nudr-dr/v2/subscription-data/{ueId}/context-data/ee-subscriptions, causing the identifier to pass validation and enter the normal create or query flow. The advisory does not independently establish whether authentication is required. Invalid identifiers can be persisted and retrieved as subscription records, enabling unauthorized data creation, UDR namespace pollution, persistent subscriber-metadata corruption, and interference with downstream components that trust stored identifiers to use valid 3GPP formats. The exposure is most relevant in lab, test, or loosely segmented service-based-interface deployments where direct UDR access is possible.22hCVE-2026-571389.9 CRI49.5%
——15PraisonAI is a multi-agent teams system. From 1.4.0 until 1.7.2, codeMode in src/praisonai-ts/src/tools/builtins/code-mode.ts executes untrusted JavaScript with new Function() inside with(sandbox) and relies on a small source-code blocklist plus shadowed process and require properties. Code can use ({}).constructor.constructor to recover the real Function constructor, obtain process and process.mainModule.require, and reach host filesystem and subprocess APIs despite the advertised sandbox. Attackers who control codeMode input can read secrets, modify files, execute commands, or exhaust the host process. This issue is fixed in version 1.7.2.1dCVE-2026-571378.8 HIG37.4%
——11PraisonAI is a multi-agent teams system. From 1.4.0 until 1.7.2, createAgentLoop() in src/praisonai-ts/src/ai/agent-loop.ts passes executable tools to generateText() before invoking the onToolCall approval callback. Because the wrapped AI SDK executes tool handlers during generation, a callback that returns false records tool_rejected only after the denied tool has already produced side effects and populated toolResults. Applications using onToolCall as a human or policy approval boundary can therefore execute rejected file, command, API, or data-modifying operations. This issue is fixed in version 1.7.2.1dCVE-2026-571368.8 HIG53.5%
——16PraisonAI is a multi-agent teams system. From 1.2.3 until 1.7.2, CommandValidator in src/praisonai-ts/src/cli/features/sandbox-executor.ts validates only the first whitespace-delimited executable against allowedCommands, then SandboxExecutor passes the complete command string to sh -c. A command beginning with an allowed executable can append a non-allowlisted command through shell metacharacters, causing arbitrary commands to run with the PraisonAI process privileges. This issue is fixed in version 1.7.2.1d