Dutch DPA Warns of OpenClaw AI Security Risks
The Dutch Data Protection Authority (AP) has issued a warning to users and organizations about the risks associated with OpenClaw and similar experimental AI systems. These open-source AI assistants, which operate autonomously on users’ computers, often fail to meet basic security standards. Users grant these assistants full access to their devices, including emails, files, and online services, allowing the AI to act independently without human approval. This level of access makes such systems highly vulnerable to misuse and cyberattacks.
Security researchers have identified that many plugins available for OpenClaw contain malware designed to steal sensitive information like login credentials and cryptocurrency assets. The platform is also susceptible to hidden commands embedded in websites, emails, or chat messages, which can lead to account takeovers, interception of personal data, and theft of access codes. Additionally, critical vulnerabilities have been found that allow attackers to remotely control entire systems, posing serious security threats.
The AP strongly advises against using OpenClaw or similar AI agents on devices containing sensitive or confidential information, such as customer files, employee data, accounting records, or identity documents. Parents are urged to monitor their children’s use of these services on home devices. Organizations should apply strict access controls, be cautious with external plugins, and renew login credentials if there is any risk of exposure. Users and organizations remain responsible for complying with the General Data Protection Regulation (GDPR) when using these technologies.
At the European level, the AP advocates for clear regulation under the AI Regulation to ensure autonomous AI agents meet safety requirements and avoid unsafe market applications. Despite running locally on users’ computers, these systems are not inherently secure. Without proper security measures, their use can result in data breaches, unauthorized access, and significant privacy violations. The responsibility to manage these risks lies with both developers and users to protect personal data effectively.
