Research
Manifold's research on agents misbehaving and the flaws they reach through, from the blog.
- Research
- Thought Leadership
From Model Risk to Action Risk
Oct 23, 2026
Oleksandr YaremchukCTO & Co-Founder
- Research
- Thought Leadership
One vendor, three platforms: governing Claude Code and Cowork
Oct 20, 2026
Guido BonomiMember of Technical Staff
- Research
The High-Risk Domains of Agent Skills and Where They Lead
Oct 13, 2026
Swapnil PatilMember of Technical Staff
Cody NashResearcher
- Thought Leadership
- Research
Four Questions to Ask Before Calling an AI Agent Rogue
Oct 7, 2026
Ax SharmaHead of Research
- Research
Ask a Model If Code Is Malicious and It Reaches for Its Morals
Oct 6, 2026- Cody NashResearcher
- Research
Jev, System One Models, and How to Use Them in Security
Oct 1, 2026- Cody NashResearcher
Markus GierlingerSecurity Researcher
- Research
Rogue Agents Targeted a Second Australian Health Dashboard, urlscan.io Index Shows
Sep 25, 2026- Ax SharmaHead of Research
- Research
Over 350,000 GitHub Files Cite Placeholder Domains Serving Scams
Sep 24, 2026- Cody NashResearcher
- Research
The third-party[.]com Domain Is Serving a ClickFix Lure to Windows Users
Sep 23, 2026- Ax SharmaHead of Research
- Research
GitSpawn: A Single Flaw Lets Untrusted Repos Run Code in Claude Code, Codex, Cursor, and Grok
Sep 1, 2026
Francisco RosalesOffensive Security Engineer
- Research
OpenAI & Hugging Face: Why the Chain-of-Thought Police Won't Save You
Aug 28, 2026- Cody NashResearcher
- Research
What to Think About curl | bash Now That AI Agents Run It.
Aug 19, 2026- Ax SharmaHead of Research
- Cody NashResearcher
- Research
Cursor CLI Ran Untrusted Repository Code With the Sandbox Switched Off
Aug 10, 2026- Francisco RosalesOffensive Security Engineer
- Research
When a Conversation Wanders, Do an Agent's Guardrails Slip?
Aug 5, 2026- Cody NashResearcher
- Research
77 "Evil Twin" Open VSX Extensions: 19 Copy Private Repo and CI Data to a New Domain
Aug 4, 2026- Ax SharmaHead of Research
- Cody NashResearcher
- Research
Anyone With an Account Could Run Commands on PewDiePie's AI Workspace (CVSS 9.9)
Jul 30, 2026- Francisco RosalesOffensive Security Engineer
- Research
The "Restricted" Deployment That Wasn't: Two Access-Control Bugs in Community-Built mcp-atlassian
Jul 27, 2026- Francisco RosalesOffensive Security Engineer
- Research
A Beaconing Counterfeit Extension on the VS Code Marketplace: The 'Markdown All Pro' Extension
Jul 22, 2026- Ax SharmaHead of Research
- Cody NashResearcher
- Research
When Your AI Reviewer Works for the Attacker: A Confused-Deputy Bug in Microsoft's Azure DevOps MCP Server
Jul 21, 2026- Francisco RosalesOffensive Security Engineer
- Research
ClaudeBleed Reopened: Browser Extensions Can Still Push Claude for Chrome to Read Your Gmail
Jul 14, 2026- Ax SharmaHead of Research
- Research
Scope Squatting: Those @openclaw and @clawhub Plugins Aren't Officially Theirs
Jun 19, 2026- Ax SharmaHead of Research
- Research
CVE-2026-52024, CVE-2026-52025: We Bypassed Two of Cline's Safety Checks to Run Code (4.2M Installs)
Jun 17, 2026- Ax SharmaHead of Research
- Research
CVE-2026-54052: A Wrong Number in n8n-mcp Leaked Your Neighbor's Credentials (CVSS 9.6)
Jun 4, 2026- Ax SharmaHead of Research
- Francisco RosalesOffensive Security Engineer
- Research
When "Read-Only Mode" Isn't: CVE-2026-46519 in mcp-server-kubernetes
May 18, 2026- Ax SharmaHead of Research
- Francisco RosalesOffensive Security Engineer
- Research
One AI Security Scanner Flagged 40% of Skills as Malicious. Most Were Fine.
May 14, 2026- Ax SharmaHead of Research
- Research
30 ClawHub Skills Are Quietly Recruiting Your AI Agent Into a Crypto Swarm
Apr 28, 2026- Ax SharmaHead of Research
- Research
An OpenClaw Skill With an SVG Can Steal Your Cookies
Apr 22, 2026- Ax SharmaHead of Research
- Research
The Gateway Gap: Why AI Agent Security Needs More Than a Chokepoint
Apr 16, 2026- Oleksandr YaremchukCTO & Co-Founder
Nate DemuthChief Architect
- Research
Two Git Commands Fooled Claude Into Merging Malicious Code
Apr 15, 2026- Ax SharmaHead of Research
- Oleksandr YaremchukCTO & Co-Founder
- Research
Non-Adversarial Agent Harm: The Overlooked Insider Threat
Apr 12, 2026
Neal SwaelensCEO & Co-Founder
- Research
Why Your EDR Has No Clue What Your Agents Are Up To
Apr 7, 2026- Nate DemuthChief Architect
- Neal SwaelensCEO & Co-Founder
- Research
Why AI Guardrails Can’t Secure Your Agents: The Classification Fallacy
Mar 31, 2026- Neal SwaelensCEO & Co-Founder
- Oleksandr YaremchukCTO & Co-Founder
- Research
The Sandbox Illusion: How Do You Isolate Agents That Think Outside the Box?
Mar 18, 2026- Neal SwaelensCEO & Co-Founder
- Oleksandr YaremchukCTO & Co-Founder
Propose research
Bring an incident or a finding: we review it and publish it with your byline.