> One fake contact
> One normal work email
> One AI agent trusted too much

Researchers showed #OpenClaw could be tricked into running hidden commands or leaking mock AWS keys, database logins, SSH credentials, and customer data.

Here's how both attacks work: https://thehackernews.com/2026/06/new-attacks-trick-openclaw-ai-agent.html

⚡ A researcher showed how three #OpenClaw flaws could have let an external #WhatsApp message trigger host code execution.

The bugs could also expose credentials or enable sandbox escape in vulnerable configurations.

Read the full story on THN 🠒 https://thehackernews.com/2026/07/researcher-details-whatsapp-to-host.html

🛑 [New] One email could rewrite what an inbox-reading AI assistant remembers about you.

MemGhost adds false information to an AI agent’s memory and changes what it says later. In #OpenClaw lab tests, the change could stay hidden from the user.

How it works, and what limits it: https://thehackernews.com/2026/07/new-memghost-attack-plants-persistent.html

‼️ Nobody asked #Claude to cancel another gym member's booking.

Aikido rebuilt the Australian gym app and tested Opus 4.6 on #OpenClaw. The model bypassed its browser-only 7-day booking limit in 9 of 10 runs.

In two runs, it also canceled another member’s reservation.

Read: https://thehackernews.com/2026/08/claude-opus-46-bypasses-gym-booking.html