Adrian Sanabria

Adrian Sanabria

@sawaba.bsky.social

🎙️ Enterprise Security Weekly Podcast Host, 🤝 BSides Knoxville Founder, 🗣️ IANS Research Faculty, 🍳 Cooking, 🏎️ F1, ⛰️ Hiking

720 followers267 following417 posts

Adrian Sanabria reposted

If you wait for KEV on this one, you're not patching - you're doing an incident response.

breachplease-fm.bsky.social@breachplease-fm.bsky.social

Two critical PHP RCEs are patching today. Jess and guest host Richy (filling in for Jake) on why 'wait for the KEV' is bad advice if you're internet-facing, and how to actually find PHP hiding in your environment. @malwarejake.bsky.social @secitup.bsky.social Watch: youtu.be/ocIIAfLlfIM

"According to documents obtained by Danish broadcaster TV 2, the attackers began accessing CPR on September 10 and executed approximately 14 million automated queries by September 20." Back in 2004, one of the things that just made sense was to graph event data and investigate volumetric spikes.

I don't think the term 'detection engineering' existed yet, and none of us knew what we were doing, but this approach was useful. Even when it wasn't malicious, answering the question, "why the spike" would teach us something useful about users/environment/business practices.

One of the most common insights from years of analyzing breaches is that there are often signs that are easy to spot, if you're looking for sudden anomalies in traffic/query/whatever volume. And defenders often have days or weeks to spot them and do something about them.

Adrian Sanabria reposted

Adrian Sanabria reposted

Booz Allen is the latest to get into AI cybersecurity benchmarks. I checked them out and share my thoughts. Sanity check: benchmarks != real world capabilities, but they do give us a data point we can use to measure AI model changes over time. defendersinitiative.substack.com/p/theres-now...

defendersinitiative.substack.comThere’s now a Cyber Weapon IndexApparently AI models are weapons now and they’re going to kill us all (but that’s a different post)

Adrian Sanabria reposted

Two things at play - ClickFix (website asks user to press Windows key + R, run a command) is *incredibly successful*, and just phoning the helpdesk and asking for password reset. The GenAI research stuff being pushed out by security and AI vendors is completely absent in the real world trenches.

Stefano Zanero@raistolo.bsky.social

Cyber insurer (Chatham house forbids me from saying who) at BH CISO summit: “so far this year 0 payouts for claims even remotely connected to a AI-driven breach, 85% of payouts related to social engineering”. Sobering statistic.

Adrian Sanabria reposted

We are on break for lunch, but rewind to see my opener… and stay tumed for 2hrs on water next. Quite timely.

IST’s #UnDisruptable27 Initiative is at #BSidesLV! @joshcorman.bsky.social leads the #IAmtheCavalry track. Tune in today to watch water regulators, engineers, and ICS experts tackle issues of operating a hospital without water, IT/OT in the AI era, #CyberCivilDefense, and more! 🛡️ Watch live:

Adrian Sanabria reposted

Normalize making fun of people when they’re all proud of having used generative AI to do some trvial task. Yesterday’s example: A colleague made a sign that said ‘USE THIS DOOR, OTHER DOOR IS BROKEN.’ Yup. AI for that. 1/?

Happy to share that I will be speaking at BSides Las Vegas this year! My talk is "Destroyed by Breach: Corporate casualties of cybersecurity failures" Common Ground, Tuesday 10:00-11:00, Florentine F I'll be at Black Hat and DEF CON as well. Let me know if you want to meet up!

This week on ESW: 1. O'Shea Bowens - network security for MCP 2. Jeremiah Grossman - there is no vulnpocalypse 3. Finally, we discuss the future of AI model use and the HuggingFace breach We went a little long on this one, but I think it was well worth it. www.youtube.com/watch?v=Em3F...

youtube.comExploring AI Network Protocols; Vulnerability Truths and Guarantees; and the News - ESW #469Segment 1 - Interview with O'Shea Bowens What do we really know about "AI Network Protocols"? Network security is about to get popular all over again. Generative AI caused a disruptive…

TL;DR - Reddit posts are now SEO/AEO, which means Reddit has become a battleground trying to fight off corporate marketing posts

Jason Koebler@jasonkoebler.bsky.social

The biohackers subreddit banned all posts about peptides and HRT because peptide companies were spamming + manipulating the sub in order to manipulate Google AI and ChatGPT answers. Very insidious industry that is drastically changing the internet www.404media.co/companies-ar...

April 2026: LOL, enterprises would have to be mad to consider using OpenClaw internally May 2026: Oh, a client has a question about some OpenClaw vulns that affect them, this is normal

You know, for an OS that sets the standard for vulnerable implementations, at least the default lock screen image is nice to look at, right?

VNC Lottery@vnclotto.misentropic.com

ASN: AS270353 Tyna Host - Datacenter no Brasil Country: BR City: São Paulo State: São Paulo Scanned: 2025-05-08T22:33:10 https://computernewb.com/vncresolver/browse#id/69826466 #vnc #infosec

Adrian Sanabria reposted

This isn't an AI problem. In any context, if the reporter didn't observe the quote, they needed to cite their source. And that's more than just the wording of the quote. The reporter is responsible for the inferred meaning matching the actual context of a statement.

Michelle Cyca@michellecyca.com

personally I think it's a very big deal that the Canada bureau chief for the @nytimes.com — certainly one of the highest-paid journalists in the country — asked an unspecified "AI tool" what Poilievre said & published its AI-hallucinated quotes in her reporting.

Adrian Sanabria reposted

Copy Fail? Also looks like memory safety, but is actually complexity. xint.io/blog/copy-fa... I am going to link to this while rejecting changes to Go crypto for years. Anyway, feeling pretty validated in my "ssh in as root, only gVisor or Firecracker are an actual security boundary" approach.

xint.ioCopy Fail: 732 Bytes to Root on Every Major Linux Distribution - XintXint Code disclosed CVE-2026-31431, an authencesn scratch-write bug chaining AF_ALG + splice() into a 4-byte page cache write. A 732-byte PoC gets root on Ubuntu, Amazon Linux, RHEL, SUSE. | AI for Se...
Filippo Valsorda@filippo.abyssdomain.expert

There are only two bug classes left: complexity and memory safety. CurveBall (CVE-2020-0601)? Complexity. BigSig (CVE-2021-43527)? Memory safety. Log4Shell (CVE-2021-44228)? Complexity. BlueKeep (CVE-2019-0708)? Memory safety. Heartbleed looks like memory safety, but it's actually complexity.

Adrian Sanabria reposted

We are very happy that today Apple issued a patch and a security advisory. This comes following 404 Media reporting that the FBI accessed Signal message notification content via iOS despite the app being deleted.