Cloud Security Podcast by Google #CloudSecPodcast #GoogleCloud #CloudSecurity Website: cloud.withgoogle.com/cloudse… Google Podcasts: podcasts.google.com/feed/aHR… Apple Podcasts / iTunes: podcasts.apple.com/us/podcas… Spotify: open.spotify.com/show/12WPC7… Twitter: x.com/CloudSecPodcast

Sunday thoughts: So you mean to tell me the managers that are so convinced they are going to replace organizations with agents are the same people saying people need to be in office. Got it.

Sharing this job position for two reasons. The first is to help a new comer find employment. The second is to point to the fact that organizations with internal access to frontier model capabilities are and will continue to hire junior security personnel. google.com/about/careers/app…

Stop Building a 2003 #SOC with #AI >> A Modern People & Process Framework bit.ly/3SrVHl9 @Medium @Anton_Chuvakin #tech #AgenticAI #leadership #management #business #CIO #CTO #CEO #CISO #cybersecurity #secops #automation #securitytools #securityoperations #cyberthreats

Companies should randomly route a small percentage of their internal AI requests to an intentionally dumb or malicious model. You want to build systems that are tolerant to different levels of intelligence and reliability.

We are leaving the Old Code Age, the Paleocodic, the artisinal code era, where if you needed a novel program, you would commission a local codesmith or code guild to hand-craft a work of code for you, bespoke.

Two simultaneous AI narratives right now: 1) You can now do the work of 20 people, learn anything and create anything with AI. Just learn how to use claude. 2) We are investing Billions in forward deployed engineers to help you implement AI b/c its too time consuming and hard for you to do it yourself. I skew more towards #1, fwiw. But I find these narratives to be incongruent.

Stop worrying about the Mythos AI cyber apocalypse. The world is more complex than “if AI is good at hacking then everyone gets hacked.”

If a vuln is publicly exploited and ends up in CISA’s KEV catalog, patching is only half the job!! We also need to know how to determine whether a server has already been compromised I’m not asking for exploit details or a PoC. I’m asking for defender guidance: - Relevant log entries - Observable artifacts - Child processes - Event IDs - File system traces Without that, orgs can patch a server that was already compromised yesterday and have no reliable way to tell Microsoft and CISA should make “Defender Detection Guidance” a standard part of every actively exploited vulnerability disclosure

My current stance on AI-assisted coding: Web pages, styling and visual templates, POCs, demos, internal tools, internal services, automation/workflows*, code reviews, test cases*, performance tuning. In short: anything that doesn’t hurt much if it breaks. YES. Standard software development, low-level programming, or coding tasks that haven’t been done publicly a thousand times and therefore are unlikely to be in the training data. NO.

"From Cloud to Chaos: Defining Shared Responsibility for AI Security" medium.com/anton-on-security… <- jokes, pain, questions, and some answers! #CoSAI

AI implementation advice on my X feed is divided between those who "feel the exponential" and those whose (unconscious?) mental model of AI is that this is about as good as it is going to get, so it is time to build around the limitations & cost structures of today's capabilities

Anthropic releases Fable with all kinds of comments about how it won’t do cybersecurity activities Meanwhile, everyone keeps trying to use it for such and complains.

This is awesome nice one @anton_chuvakin sharing CoSAI fundamentals as we build for AI security The framework turns ‘whose fault is this?’ into ‘which layer’s controls failed, and who owns remediation for each?

The current lolz is that everyone is AI assisted REing EDR products, backchanneling stories and findings, and quietly comparing notes. People are just avoiding talking about it publicly so vendors don’t start harassing them, targeting tenants, or doing other fun things.

Please don't forget about other cyber threats.

"From Cloud to Chaos: Defining Shared Responsibility for AI Security" bit.ly/4oXmZf4 <- jokes and painful questions (and some answers) from CoSAI work on AI shared responsivity

it happens