Find notable cyber news and cases, enriched with sources, timelines, and signals.
Home Watchlists Beta Browse News Feed Stats About Contact

OpenClaw message-object prompt injection patched in 2026.4.23 security flaw

Vulnerability
First reported
Last updated
Happening score
H score 15
1 unique sources, 1 articles

Summary

Hide ▲

OpenClaw has a patched message-object prompt injection flaw that let hidden instructions inside shared contacts, vCards, and location pins reach the LLM as trusted prompt text, creating code-execution and secret-leakage risk for self-hosted agents. Imperva showed the agent flattening those objects inline with no untrusted boundary, so attacker text could blend into ordinary-looking input. OpenClaw 2026.4.23 fixes the issue by moving contact names, vCard fields, and location labels into a separate untrusted-metadata channel. Systems that have not updated remain exposed whenever they ingest untrusted message content.

Related Happenings

MemGhost stealth memory injection against OpenClaw personal agents

Technical Analysis
H score23 First: 13.07.2026 16:49 Last: 13.07.2026 16:49 Sources 1

About this happening: Researchers demonstrated MemGhost, a one-email prompt-injection technique that can plant a persistent false memory in OpenClaw-style personal agents, letting an at...

Open Happening

OpenClaw command-injection, path-traversal, and link-following flaws (multiple vulnerabilities)

Vulnerability
H score33 First: 10.07.2026 17:19 Last: 10.07.2026 17:19 Sources 1

About this happening: OpenClaw now has three patched high-severity vulnerabilities that can lead to credential theft, privilege escalation, and arbitrary code execution on the host....

Open Happening

OpenClaw outbound-mail approval gates and trust-scoped connector controls

Defensive Guidance
H score11 First: 11.06.2026 20:46 Last: 11.06.2026 20:46 Sources 1

How related: Outbound mail needs a gate: no first-time sends to unfamiliar addresses without approval, so a hijacked agent cannot relay phishing from a trusted account.

About this happening: OpenClaw operators are adding outbound-mail approval gates, trust-scoped connector access, and human approval for risky actions to reduce agent phishing and unauth...

Open Happening

OpenClaw/OpenShell managed sandbox backend Claw Chain (multiple vulnerabilities)

Vulnerability
H score31 First: 15.05.2026 16:35 Last: 15.05.2026 16:35 Sources 1

About this happening: Researchers disclosed four OpenClaw flaws in the OpenShell managed sandbox backend that can be chained for data theft, privilege escalation, and persistence. T...

Open Happening

ChatGPT single-prompt DNS side-channel exfiltration remote code execution flaw

Vulnerability
H score33 First: 31.03.2026 16:01 Last: 31.03.2026 16:01 Sources 1

About this happening: A ChatGPT vulnerability let a single malicious prompt covertly exfiltrate prompts, messages, uploaded files, and other sensitive content through a DNS side channel. Th...

Open Happening

Timeline

  1. 11.06.2026 20:46 2 articles · 1mo ago

    OpenClaw flaws let hidden instructions run code and phishing emails leak secrets

    Initial Disclosure

    Imperva and Varonis Threat Labs disclosed separate findings showing that OpenClaw can be manipulated through ordinary-looking inputs. Imperva demonstrated that hidden instructions embedded in shared contacts, vCards, and location pins could be flattened into prompt text and made the agent download and run a script, while Varonis showed that a Pinchy agent could be tricked by a believable email into forwarding mock AWS IAM access keys, database connection strings, SSH credentials, and a synthetic customer export.

    Show sources
    Open in new tab