My analysis from hosting Hugging Face at the CISO huddle for the Cloud Security Alliance, operational to program building. Remember: Agents… find a way. Thanks: Sergej Epp - huge support, Sri… | Gadi Evron | 43 comments

My analysis from hosting Hugging Face at the CISO huddle for the Cloud Security Alliance, operational to program building. Remember: Agents… find a way. Thanks: Sergej Epp - huge support, Sri Srinivasan, Rich Mogull, Rob T. Lee, Jim Reavis, Sounil Yu for collaboration. And thank you Hugging Face. My personal ask: If you need to discover and defend agents, do keep Knostic and myself in mind. Observations on dealing with an autonomous AI adversary: 1. Purely task-focused, with no observed motivation 2. A bias to repeating the same attempts once successful 3. Brilliant attacks followed by basic actions 4. High-speed operations run simultaneously 5. Taking paths no human would take 6. Classic attacks, with a focus on package manager vulnerabilities, AppSec flaws, and credentials theft 7. Benchmark strings throughout the traces 8. Hallucinated output at scale, and log comments reading like agent reasoning Detection difficulties: 1. Many paths and techniques at once + signal indistinguishable from noise 2. Traditional systems designed for 1-2 attack paths, now seeing many vectors 3. Systems triggered alerts but at wrong criticality level. 4. Attendee considerations: Deception would have helped. Inference speed is a limitation, consider classifiers for triage. Systemic lessons: 1. Using coding agents is the new reality, and without, response would have taken weeks 2. You can't build a defense program without open weight models. Hugging Face hit guardrails on Opus, Fable 3. Legitimate agentic platform usage resembles attack patterns. Immediate response lessons: 1. Be able to mass-rotate all credentials and secrets 2. To destroy and rebuild clusters 3. Rapid custom UIs generation with AI overshadowed security tools 4. AI timeline reconstruction + hunting for deeper compromise as core capabilities 5. Collaboration through shared, annotated events UI My strategic/security program takeaways: 1. The new AI basics: - Instrument agents to extend your security detection and response/SPM stack into the agents themselves - Use deception tech to slow down attackers A sandbox doesn't cut it. Classic basics and permissions, a good practice, *won't* be effective against an agent. 2. Strategic: without open weight models, you can't reliably defend yourself. Being able to shift models at will, when lab models refuse cyber queries, is critical. 3. Logistical: reserve a token budget. Incident response has a cost, which includes a significant token budget. Critically, the same is true for the attacker's side, estimated at $100K here. 4. Operational: prepare for hallucinated artifacts in detection and forensics, at scale. Dealing with forensic traces left behind by the model wastes endless defender cycles. All the lessons of the past hold, just at a new scale. We will deal with a tsunami of indistinguishable findings, all at once. Or put another way, being attacked by a thousand "soldiers" at once, even if they aren't too intelligent, is overwhelming. | 43 comments on LinkedIn

linkedin.com

Protect Your Privacy with California's DROP Tool

Are you a California resident? Then we've got exciting news for you: there's a tool just for you that lets you take a single, relatively easy step to protect your privacy. It's called a DROP request. ...

eff.org

Finding vulnerabilities was never the hard part

AI is flooding firms with security data. Fortress Security CEO Alex Santos warns that cybersecurity vulnerability prioritization—not discovery—is the real challenge.

cyberscoop.com

Flipper Zero firmware development continues with community help

Flipper Devices says development of the Flipper Zero firmware will continue, albeit with a smaller internal team and greater reliance on community contributions.

bleepingcomputer.com

Victory! Supreme Court Says Constitution Protects People’s Location Data

You have an expectation of privacy in location data that reveals your movements in the physical world, and even short-term surveillance of these movements is a search subject to the Fourth Amendment, ...

eff.org

The Biggest Attack on Abortion Since the End of Roe

A federal court wants to end mifepristone by mail—it's not happening

jessica.substack.com

about the lab – cognitive tools lab

cogtoolslab.github.io

Eulogy for the CIA Factbook: The free standard for world facts, long an educational staple, is gone

The Trump administration has shut down the CIA World Factbook, and there's much lamenting about the demise of a free source that many people used to check basic facts about countries.

apnews.com

AI is teaching us to speak like bots and it’s a problem

AI is creeping into our everyday conversations, making us less patient and teaching us to prompt others instead of talking to them.

fastcompany.com

A top AI researcher explains the limitations of current models

François Chollet talks about his deceptively simple new benchmark test for AI models.

fastcompany.com

Prof. Judy Fan: Cognitive Tools for Making the Invisible Visible

YouTube video by MIT Siegel Family Quest for Intelligence

youtu.be

Emergent Cyber Behavior: When AI Agents Become Offensive Threat Actors - Irregular

In controlled experiments, AI agents performing routine enterprise tasks were found to autonomously engage in offensive cyber operations, including vulnerability exploitation, privilege escalation, an...

irregular.com