OpenClaw AI Assistant Compromised via WhatsApp: Three Critical Vulnerabilities Allowed Credential Theft, Sandbox Escape, and Arbitrary Code Execution on Host
Security researchers have disclosed three serious vulnerabilities in OpenClaw that could allow attackers to steal credentials, escalate privileges, and execute arbitrary code on systems running the AI assistant. The flaws affected both home users and corporate deployments, highlighting risks associated with insufficient input validation and sandbox isolation in AI tooling.
Command Filtering Failures (CVSS 8.8)
Two of the vulnerabilities received a CVSS score of 8.8. They were caused by incomplete command filtering mechanisms that did not block all malicious inputs. As a result, an attacker could inject system commands into contexts where they should have been prohibited, enabling unauthorized actions ranging from data exfiltration to privilege escalation.
Sandbox Bypass via Directory Mounting (CVSS 8.4)
The third vulnerability, rated 8.4, allowed attackers to circumvent sandbox restrictions by mounting parent directories. Although OpenClaw blocked direct access to sensitive folders such as ~/.ssh, ~/.aws, and ~/.gnupg, it could still permit mounting of the entire /home directory. This effectively placed SSH keys, cloud tokens, and GPG secrets directly accessible to the attacker.
Further escalation was possible through the /var directory, which could grant access to the Docker socket and enable escape from the sandbox directly onto the host system. Researcher Chinmohan Nayak demonstrated that the attack could be initiated remotely via an external message sent through WhatsApp (owned by Meta, designated as extremist and banned in Russia) without requiring any prior foothold on the target machine.
Impact and Recommendations
In the worst-case scenario, an attacker could exfiltrate sensitive data, establish persistence, and execute arbitrary code on the victim’s machine. Developers addressed all three issues in OpenClaw 2026.6.6. Users are strongly advised to:
- Update to the latest version immediately
- Enable sandboxing for all secondary sessions
- Remove exec from the list of allowed tools
- Narrow the list of trusted communication channels to the minimum necessary
These measures significantly reduce the attack surface and help prevent similar exploitation attempts in the future.
Related articles
Critical Vulnerability in JetBrains TeamCity Allows Unauthenticated Remote Code Execution
JetBrains has disclosed a critical vulnerability in its on-premises TeamCity CI/CD server that permits remote attackers to execute arbitrary operating system commands without authentication. The flaw, tracked as CVE-2026-63077, affects the agent polling protocol and can be exploited simply by accessing the TeamCity Server over HTTP or HTTPS. With a CVSSv3.1 base score of 9.8, the issue is rated Critical and could lead to data theft, configuration changes, or compromise of build artifacts and downstream pipelines. No in-the-wild exploitation had been observed at the time of disclosure. JetBrains has released fixed versions TeamCity 2026.1.3 and 2025.11.7, along with a patch plugin for all releases since 2017.1, and recommends restricting external access until updates can be applied.
Critical OS Command Injection Flaw in Arista VeloCloud Orchestrator Exploited in the Wild
Arista Networks has disclosed a critical vulnerability in the on-premises version of its VeloCloud Orchestrator product used for centralized SD-WAN management. The flaw, tracked as CVE-2026-16812, is an OS command injection issue that allows remote attackers to compromise the system without any authentication. Exploitation has already been confirmed, and the vulnerability carries the maximum CVSS base score of 10.0 in both version 4.0 and 3.1, classifying it as Critical. Successful attacks can impact the confidentiality, integrity, and availability of the orchestrator and all managed data. In addition, a compromised VeloCloud Orchestrator instance may grant attackers access to connected VeloCloud Edge devices across the network.
CISA Adds Exploited Flaws in FortiOS and VeloCloud Orchestrator to Known Exploited Vulnerabilities Catalog
The US Cybersecurity and Infrastructure Security Agency has added two actively exploited vulnerabilities to its Known Exploited Vulnerabilities catalog. CVE-2026-16812 affects the on-premises version of Arista VeloCloud Orchestrator and allows OS command injection that can lead to data leakage, tampering, and denial of service. The flaw carries a maximum CVSS v3.1 base score of 10.0 and is rated Critical. CVE-2025-68686 impacts Fortinet FortiOS and can be used to bypass specific patches and steal sensitive information when combined with another vulnerability that grants prior filesystem access. Federal agencies must remediate the VeloCloud issue by July 30 and the FortiOS issue by August 10.
Critical Vulnerabilities in JetBrains IntelliJ IDEA and TeamCity Enable Remote Code Execution
JetBrains has addressed multiple critical vulnerabilities in its IntelliJ IDEA and TeamCity products that could allow remote code execution, unauthorized file access, and sandbox escapes. The flaws affect remote development environments and require immediate patching, especially in shared setups. In IntelliJ IDEA, CVE-2026-59792 involves directory traversal during workspace identifier processing and carries a CVSS score of up to 9.8. TeamCity received fixes for several issues, including malicious Git repository configurations that enable code execution and a Kotlin DSL sandbox escape. Additional patches cover arbitrary file access through Perforce integration and a persistent cross-site scripting flaw on cloud profile pages. Administrators are urged to update IntelliJ IDEA to versions 2026.1.4 or 2026.2 and TeamCity to 2026.1.2 or 2025.11.6 depending on the release line in use.