#ml-security · Louche Cannon

MCP for agent-to-agent comms may be the riskiest protocol you've never heard of - Ars Technica

Tue, 06 Oct 2026 09:04:14 +0200

Bookmarked https://arstechnica.com/security/2026/10/vulnerability-in-agents-from-google-and-others-exposes-structural-flaw-in-mcp/

Yann LeCun: Anthropic CEO Dario Amodei is ‘deluded,’ ‘crazy,’ and doesn’t understand cybersecurity | Fortune

Fri, 02 Oct 2026 20:38:54 +0200

Bookmarked https://fortune.com/2026/10/01/yann-lecun-anthropic-ceo-dario-amodei-deluded-crazy-cybersecurity/

Your Open Source Model Could Have a Hidden Time-Release Backdoor | Morgin.ai

Mon, 24 Aug 2026 11:44:50 +0200

Bookmarked https://morgin.ai/articles/your-open-source-model-could-have-a-hidden-time-release-backdoor.html

Everything I own, owned - schlarp.com

Mon, 24 Aug 2026 06:53:43 +0200

Bookmarked https://schlarp.com/posts/everything-i-own-owned/

Terabytes of credentials leaked in massive supply-chain attack - Ars Technica

Thu, 13 Aug 2026 08:09:08 +0200

Bookmarked https://arstechnica.com/security/2026/08/terabytes-of-credentials-leaked-in-massive-supply-chain-attack/

Docker Sandboxes | Sandboxes for Coding Agents | Docker

Thu, 13 Aug 2026 08:06:04 +0200

Bookmarked https://www.docker.com/products/docker-sandboxes/

Anthropic says it will watermark text generated by its AI models | TechCrunch

Wed, 12 Aug 2026 09:33:08 +0200

Bookmarked https://techcrunch.com/2026/08/11/anthropic-says-it-will-watermark-text-generated-by-its-ai-models/

Anthropic's Fever Dream: Claude's package anthropickit that stole real keys

Mon, 03 Aug 2026 07:14:56 +0200

Bookmarked https://www.aikido.dev/blog/anthropic-rogue-agents-package-stole-keys

OpenAI says its AI went rogue and launched 'unprecedented' cyber-attack

Wed, 22 Jul 2026 15:04:48 +0200

Bookmarked https://www.bbc.com/news/articles/c3ek3gvdnj3o

The Memory Heist

Wed, 15 Jul 2026 14:16:05 +0200

Bookmarked https://www.ayush.digital/blog/the-memory-heist

Prompt Injection as Role Confusion

Thu, 25 Jun 2026 13:28:29 +0200

Bookmarked https://simonwillison.net/2026/Jun/22/prompt-injection-as-role-confusion/#atom-everything

Feds freaked over Fable 5 after simple 'fix this code' prompt, not jailbreak, says researcher

Tue, 16 Jun 2026 22:02:22 +0200

Bookmarked https://www.theregister.com/security/2026/06/15/feds-freaked-over-fable-5-after-simple-fix-this-code-prompt-not-jailbreak-says-researcher/5255827

‘The Worst Leak That I’ve Witnessed’: U.S. Cybersecurity Agency Leaves Its Digital Keys Out in Public on GitHub

Tue, 19 May 2026 21:56:21 +0200

Bookmarked https://gizmodo.com/the-worst-leak-that-ive-witnessed-u-s-cybersecurity-agency-leaves-its-digital-keys-out-in-public-on-github-2000760330

Dependency cooldowns turn you into a free-rider

Wed, 15 Apr 2026 07:34:49 +0200

Bookmarked https://calpaterson.com/deps.html

[2604.04757] Undetectable Conversations Between AI Agents via Pseudorandom Noise-Resilient Key Exchange

Wed, 08 Apr 2026 19:23:12 +0200

Bookmarked https://arxiv.org/abs/2604.04757

"Disregard that!" attacks

Thu, 26 Mar 2026 09:00:04 +0200

Bookmarked https://calpaterson.com/disregard.html

From magic to malware: How OpenClaw's agent skills become an attack surface | 1Password

Thu, 05 Feb 2026 19:20:11 +0200

Bookmarked https://1password.com/blog/from-magic-to-malware-how-openclaws-agent-skills-become-an-attack-surface

Sandboxing AI agents in Linux — Senko Rašić

Wed, 04 Feb 2026 08:52:07 +0200

Bookmarked https://blog.senko.net/sandboxing-ai-agents-in-linux

OpenClaw (a.k.a. Moltbot) is Everywhere All at Once, and a Disaster Waiting to Happen – Communications of the ACM

Wed, 04 Feb 2026 08:45:49 +0200

Bookmarked https://cacm.acm.org/blogcacm/openclaw-a-k-a-moltbot-is-everywhere-all-at-once-and-a-disaster-waiting-to-happen/

Self-driving cars, drones hijacked by custom road signs • The Register

Sun, 01 Feb 2026 09:12:24 +0200

Bookmarked https://www.theregister.com/2026/01/30/road_sign_hijack_ai/