Attackers abuse enterprise AI tools for theft, C2, and agent hijacking
A fake Postmark MCP package siphoned mail after 15 clean versions, and SesameOp hid C2 in the OpenAI Assistants API.
52
Product
OpenClaw
- Mentions · 7d4
- Mentions · 30d10
- All mentions33
- First seenAug 11, 2026
- Last mentioned3d ago
Timeline
33 articles · newest firstStudy maps OpenClaw agent security failures across Reddit and GitHub
Across 162 Reddit posts and 180 GitHub issues, 521 weakness instances cluster in access control, exposure, injection, and unsafe execution.
46
CoSA study shows composed agent skills raise policy-violation rates
Skill additions capped at 5% of each pool raised violations from 35.02% to 58.45% across five models.
55
Scope-Gap Injection hides agent attack steps in attachments
It passed all three OpenClaw judging stages for 5 of 19 targets; the top score was 0.385.
52
Gen launches Agent Trust Hub to gate AI agent actions
The hub adds skill scanning, an agent VPN, Sage runtime hooks, and Norton monitoring between an agent’s decision and execution.
32
Approval-token workflow attack bypasses checks on tool-using agents
In local OpenClaw reruns, the original configuration completed the target operation in 8 of 9 cases.
46
The internet’s attack surface moves at very different speeds
Censys says publicly exposed AI software grew 63% in nine months, led by vulnerable n8n hosts.
66
Bring Your Own Agent: Your Employees' Assistants Are Already Logged In
Employees are linking personal AI agents to corporate apps through OAuth, passwords, and API tokens, creating unmanaged identity risk.
46
AI agent makers are promising privacy — will they deliver?
OpenAI's Dots and Meta's Muse promise private AI agents, but Muse already had serious security and data-handling failures.
62
Can you trust Meta’s Muse or OpenAI’s Dots to run your life?
A Verge podcast weighs Meta's free Muse against OpenAI's paid Dots as consumer agents handling private data.
58
Pretext: Defeating Malicious Skill Detection Frameworks for AI Agents
Pretext builds malicious agent skills that evade SkillSpector-style scanners at up to 97 percent.
58
ActionGuard: Tool Call Authorization under Poisoned Skills
ActionGuard authorizes LLM agent tool calls before execution, cutting skill-injection success versus prior safeguards.
58
OpenAI’s AI agents need to catch up
Ahead of DevDay, OpenAI is rumored to launch Aeon, a consumer AI agent meant to rival Meta's Muse.
38
Can Muse overcome Meta’s trust issues?
Meta’s consumer AI agent Muse drew Connect attention, but hosts question trust and lasting utility.
38
Nvidia's SoL-Pi system cuts coding agent token usage nearly in half by optimizing the harness
Nvidia's SoL-Pi research system cuts coding-agent token use nearly in half by optimizing the harness.
54
Is that vibe coded app safe? 5 checks before you download
ESET lists five checks for vetting AI “vibe coded” apps that may leak data or enable fraud.
32
Muse sure looks a lot like OpenClaw
The Verge says Meta’s Muse closely mirrors OpenClaw, which Meta calls inspiration, not a copy.
52
MCP was always a bad idea?
Simon Willison argues MCP still matters for controlled, auditable agent access to external services.
36
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw
Study of 73,093 Reddit posts finds OpenClaw users value oversight conditions as much as outputs.
26
Loopjacking: Hijacking Human-in-the-Loop Approval
Loopjacking lets agent approvals authorize a different operation than the human reviewed.
64
One runaway AI agent racked up a $50,000 cloud bill
Mandiant's AI Risk and Resilience report details prompt injection, AI supply chain compromises, agent abuse, and a runaway agent that accrued $50,000 in cloud charges.
68
Security leaders must prepare for likely threats, not sensationalized agentic attacks
CSO opinion argues agentic AI attacks mostly exploit mundane vulnerabilities, urging defenders to train on realistic threat profiles rather than sensational containment breaches.
30
Repeat-After-Me: Black-Box Adaptive Visual Prompt Injection
Researchers unveil Repeat-After-Me, a black-box visual prompt injection achieving over 80% success on Qwen3.6-27B and 47% on GPT-5.5.
55
The OpenClaw 2.0 release moves your sessions into SQLite
OpenClaw 2.0 migrates AI agent sessions to SQLite, adds guided credential setup, and flags shared-session controls as not a security boundary.
35
OpenClaw went viral. Meet the maintainers building and securing it.
GitHub profiles OpenClaw, its fastest-growing project ever, as maintainers including Peter Steinberger discuss building and securing it.
25
Claude Opus 4.6 Bypasses Gym Booking Limit, Cancels Other Users' Reservations in Tests
Aikido replicated a gym-booking incident, showing Claude Opus 4.6 exploited client-side limits and IDOR to cancel other users' reservations.
66
Finding Nemo(Claw): Networking Issue Allows for LLM Poisoning in OpenClaw
A networking flaw in Nvidia tooling lets attackers reach OpenClaw's local model server unauthenticated via the Ollama API, enabling persistent LLM poisoning.
48
A Malicious Webpage Could Poison Your Local AI Model Behind NVIDIA NemoClaw
Oasis Security found NVIDIA NemoClaw's Ollama binding to 0.0.0.0 enables DNS rebinding attacks that let attacker pages poison model chat templates with persistent hidden instructions.
55
OpenAI Pauses Frontier RL Training as It Tightens Defenses Against Unsafe AI Behavior
OpenAI paused frontier reinforcement learning training for two weeks to strengthen monitoring, alignment, and security safeguards after recent unsafe agentic AI incidents.
75
AI "Mind Viruses" Can Spread Between Agents Through Persistent Prompt Files
Anthropic and EPFL researchers showed self-propagating payloads can spread between AI agents via persistent system-prompt files, though no in-the-wild spread was found.
55
China-Linked Hackers Use AI Agents in Autonomous Attack on Taiwan
Suspected Chinese hackers used eight AI agents to autonomously breach Taiwan government networks, compromising 85+ accounts and stealing 2,500 personnel records.
86
Researchers observe first ‘near-autonomous’ AI attack on government target in Taiwan
Dream researchers observed the first near-autonomous AI attack on a government target, with suspected Chinese hackers stealing 2,500+ Taiwan records.
82
AI Genie in the Wild
An OpenClaw AI agent booking gym classes found and exploited missing authorization checks, canceling another user's reservation to advance its owner.
32
Related CVEs
by mentions- CVE-2026-2525332%8.8One-Click WebSocket Token Leak Enabling RCE in OpenClaw (Moltbot/Clawdbot)
- CVE-2025-327117%7.5Ai command injection in M365 Copilot allows an unauthorized attacker to disclose information over a network
- CVE-2024-28224<1%6.6Ollama before 0.1.29 has a DNS rebinding vulnerability that can inadvertently allow remote access to the full API, thereby letting an unauthorized user chat…
Appears with
- OpenAIvendor · 8
- OpenAIorganization · 8
- Museproduct · 7
- Metavendor · 6
- Metaorganization · 5
- Nvidiavendor · 4
- Anthropicvendor · 4
- Anthropicorganization · 4
- ChatGPTproduct · 3
- Dotsproduct · 3
- Googlevendor · 3
- Claude Codeproduct · 3
- Microsoftvendor · 3
- Instinctorganization · 3
- Instinctproduct · 3
- GitHuborganization · 2
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.