OpenClaw gives users yet another reason to be freaked out about security



For more than a month, security practitioners have been warning about the perils of using OpenClaw, the viral AI agentic tool that has taken the development community by storm. A recently fixed vulnerability provides an object lesson for why.

OpenClaw, which was introduced in November and now boasts 347,000 stars on Github, by design takes control of a user’s computer and interacts with other apps and platforms to assist with a host of tasks, including organizing files, doing research, and shopping online. To be useful, it needs access—and lots of it—to as many resources as possible. Telegram, Discord, Slack, local and shared network files, accounts, and logged in sessions are only some of the intended resources. Once the access is given, OpenClaw is designed to act precisely as the user would, with the same broad permissions and capabilities.

Severe impact

Earlier this week, OpenClaw developers released security patches for three high-severity vulnerabilities. The severity rating of one in particular, CVE-2026-33579, is rated from 8.1 to 9.8 out of a possible 10 depending on the metric used—and for good reason. It allows anyone with pairing privileges (the lowest-level permission) to gain administrative status. With that, the attacker has control of whatever resources the OpenClaw instance does.

“The practical impact is severe,” researchers from AI app-builder Blink wrote. “An attacker who already holds operator.pairing scope—the lowest meaningful permission in an OpenClaw deployment—can silently approve device pairing requests that ask for operator.admin scope. Once that approval goes through, the attacking device holds full administrative access to the OpenClaw instance. No secondary exploit is needed. No user interaction is required beyond the initial pairing step.”

The post continued: “For organizations running OpenClaw as a company-wide AI agent platform, a compromised operator.admin device can read all connected data sources, exfiltrate credentials stored in the agent’s skill environment, execute arbitrary tool calls, and pivot to other connected services. The word ‘privilege escalation’ undersells this: the outcome is full instance takeover.”



Source link

  • Related Posts

    Netflix launches a standalone app for kids’ games

    Netflix is launching a new standalone app for kids’ games called Netflix Playground, the company announced on Monday. Netflix Playground is available as part of a Netflix subscription, and doesn’t…

    Three YouTubers accuse Apple of illegal scraping to train its AI models

    Three YouTube channels have banded together and filed a class action lawsuit against Apple, as first spotted by MacRumors. According to the lawsuit, the creators behind h3h3 Productions, MrShortGameGolf and…

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    You Missed

    In the age of autocorrect, this typewriter shop owner in Hamilton wants you to write with intention

    In the age of autocorrect, this typewriter shop owner in Hamilton wants you to write with intention

    Wow Nepal – Marginal REVOLUTION

    Wow Nepal – Marginal REVOLUTION

    Mapping the scale of the conflict.

    I Prefer hotels to book with Citi points before devaluation

    I Prefer hotels to book with Citi points before devaluation

    Sporting vs. Arsenal prediction, odds, best bets for UCL quarterfinal first leg

    Sporting vs. Arsenal prediction, odds, best bets for UCL quarterfinal first leg

    Canada Gazette – Part I, May 6, 2023, volume 157, number 18