
CIO & CISO @Wiz_io - now part of @GoogleCloud! Previously sec eng @ Meta, CTO @ Tanium, IR director @ Mandiant, and consultant for #MrRobot. Opinions my own.
DC Metro
Joined July 2009
- Tweets335
- Following379
- Followers4.8K
- Likes2.2K
We seeded 2m canaries into popular AI services and tracked where they were stolen and abused.
This identified:
- Three compromised providers
- Resellers stealing prompts via built-in logging
- Providers ignoring ZDR & privacy flags
dragos.com/blog/ai-supply-ch…
Ryan Kazanciyan retweeted
Just got the email from Openrouter and now the story is confirmed
Atlascloud, Novita (who I know was flagged US with ZDR on openrouter) and Streamlake were… oops! Forwarding your data BACK TO MINIMAX IN CHINA
I'm removing OpenRouter and Vercel Gateway from my stack as a result of this. No specific threat, just not worth the risk. I'm moving all inference not from hyperscalers/frontier labs to Baseten
github.com/nateberkopec/dotf…
Ryan Kazanciyan retweeted
Some commentary on autonomous offensive frameworks
trustedsec.com/blog/accuracy…
Ryan Kazanciyan retweeted
I very much appreciate all the organizations sharing their IOCs and lessons from the recent "Black October" spree of breaches. It can be easy to bury your head in the sand, but we can all learn from these incidents
Here's one I haven't seen discussed: Ardor Cloud
Ryan Kazanciyan retweeted
New AWS feature for helping with least privilege: scoped credentials. Thankfully AWS has also released a way to debug access denieds.
aws.amazon.com/about-aws/wha…
aws.amazon.com/about-aws/wha…
Ryan Kazanciyan retweeted
US Was Warned of Telecoms Risk Years Before Salt Typhoon Hack
Good get by @rj_gallagher and Bloomberg to dislodge this 2022 CISA report, which took a FOIA lawsuit to finally obtain.
bloomberg.com/news/articles/…
Super interesting work!
Testing browser agents like @googlechrome auto browse on the live web is tricky: sites change, side effects persist & testing prompt injection on live pages isn't safe!
So our team built a "Mirror Dimension": a hermetic, synthetic web to safely test agents! 🪞🤖
🧵
3
353
Ryan Kazanciyan retweeted
No.
1. In the presence of superhuman bug finding machines, humans cannot write safe code without extreme levels of care and planning.
2. That care has almost never been applied to the critical systems in our lives. Not because software engineers don't care, but we aren't really engineers, not in the way civil engineers are engineers. The title is stolen valor.
3. Even careful, experienced "engineers" have difficulty writing code that is safe against modern models in C/C++ and other low-level languages.
4. We are finding that CVEs are overrated; huge numbers of organizations are being owned by custom vulns in the 100s of billions of lines of non-standard enterprise code out there or misconfigurations in heterogeneous systems, which LLMs can find and write exploits for live.
5. @dinodaizovi is right, we need architectures for which one RCE is not game-over and that buy us time for AI-driven response.
6. Everything is going to have to be re-written by AI, in type/memory safe languages, with minimal libraries, using formal methods, contracts, solvers, etc...
The vulnpocalypse is caused by security vendors.
Their inability to convey risk accurately, due to either incompetence or ignorance, causes tremendous damage and stress on responding teams.
There is a lot of weight associated with vulnerability announcements. We can no longer trust many security vendors’ blog posts, since their purpose is marketing & fanfare, not accuracy.
Exploitability needs to be demonstrated, in realistic environments, and its likelihood accurately conveyed. So that responders can make informed decisions on what to prioritize. But instead we are seeing fake PoC videos set in contrived & controlled lab environments, that give an illusion of exploitability when it’s unrealistic in any real-world scenario.
And vendors now try to show off who found it first, rush to send you marketing emails and call you without your consent; to sell you snake oil they pulled out of their ass two hours earlier, with no intention on actually making anyone safer, only to meet their quarterly sales quotas.
The core responsibility of a security professional is to accurately communicate risk, and let people make informed decisions. Too many vendors are failing at this, that I cannot consider them security vendors, only snake oil salesmen taking advantage of people in vulnerable positions due to their lies.
Ryan Kazanciyan retweeted
New from HD Moore:
"We sent nearly two million AI prompts containing unique, trackable secrets to AI providers, then watched where those secrets surfaced. What we found revealed security breaches, prompt theft, and widespread failures to protect customer data." @hdmoore
dragos.com/blog/ai-supply-ch…
Ryan Kazanciyan retweeted
CVSS 9.9 in Oracle Database. Oracle just patched CVE-2026-114819, which I reported. Chained with a GraphQL bug I disclosed in October, it turns an exposed HTTP endpoint into command execution on the database host.
Then I wondered if Oracle was the exception. It wasn't. PostgreSQL and MySQL both fell to full RCE from the lowest-privilege accounts you can create.
These were meant for Pwn2Own, but we failed to register. So you get the write-up instead.
depthfirst.com/research/narr…
Ryan Kazanciyan retweeted
Today we disclose CVE-2026-123633 and CVE-2026-123007 (CVSS 4.0: 9.2) - exploitable memory corruptions in NGINX
We achieved reliable remote code execution (ASLR defeated) on NGINX serving static files - no backend or third-party modules 🤯
Both require HTTP/2
Read more 🧵⬇️
Ryan Kazanciyan retweeted
Microsoft's agent-framework hack was conducted using Cairn, Pi (harness), and GLM 5.3
The attacker compromised credentials to ESRP (MSFT release pipeline), and navigated IP allowlisting to release the malicious version
Ryan Kazanciyan retweeted
An update on Northflank's October 30 security incident.
We’ve been working with independent forensic firm FTI Consulting to investigate. We’ve now published our full report, covering what happened and the changes we’re making.
The key findings:
1/ On October 30, an attacker exploited a Northflank API vulnerability and exfiltrated secrets affecting 1.5% of customer teams, representing 1.8% of stored secret values.
2/ The investigation has found no evidence of access to customer workloads or workload data.
3/ We fixed the vulnerability within 82 minutes of receiving the first alert. By the time we posted the status update on October 30, the attacker was no longer in the environment and it was safe to rotate credentials.
4/ We’ve strengthened detection and automated blocking, and are continuing platform hardening and AI-assisted red-teaming.
I’m deeply sorry for the worry, disruption and extra work this has caused. Thank you for the patience and kindness you’ve shown our team. We know rebuilding your trust takes sustained work, and we’ll keep sharing our progress.
A huge thank you to the Wiz incident response team and Ami Luttwak for their support and expertise throughout this period.
northflank.com/blog/incident…
Ryan Kazanciyan retweeted
GTIG just put out the AI threat tracker, full of case studies from Mandiant IRs, Gemini, and everywhere else we’re watching the threat evolve.
Big takeaways for me are the n-day problem (speed), the agentic frameworks (scale), and LLMJacking (scale).
cloud.google.com/blog/topics…
Ryan Kazanciyan retweeted
This is a big one.
We found malicious JavaScript being served through Optimizely to some of the world's largest websites.
178 organisations affected, with everything from crypto theft and card skimming to iOS exploits.
Zoom, BBC, CNN, HSBC, Salesforce, GitLab, KLM, IKEA, Equifax, Ledger. The list goes on... Full list in the post.
blog.report-uri.com/worlds-l…
Ryan Kazanciyan retweeted
Every outage—cloud providers, air traffic control, you name it—brings on swarms of people looking for "gotchas" to fit whatever narrative is in their head.
We'd all be better off giving it 24-48 hours for actual facts to surface before dog piling. Wishful thinking.
Replying to @GergelyOrosz
Also there's the narrative that "Microsoft layoffs" and "no one left who understands how stuff works." All untrue.
They had dozens of parallel workstreams and mitigations, like a bee hive. More than 1,000 engineers were involved. It was just a damn tricky issue, and it triggered a bunch of red herrings as well as I understood
It's a "what could go wrong, did go wrong" example. Which IS to be expected at global scale!! So not an excuse, just context
Ryan Kazanciyan retweeted
Replying to @MalwareTechBlog
Like I said.. I don’t know. Here are some possible outcomes.
1) projects roll into a foundation (Apache, Linux, etc) that get commercial funding but maintain independence.
2) entities (industry groups) create forked repos and maintain shadow projects with commercial support. There is a growing call to do this for critical infra.
3) projects are replaced by safer alternatives that are commercially backed. I suspect this could happen with many smaller projects and their replacements can be spec-driven replaced quickly in safer languages. I’d like to see this for media libraries.
4) very critical projects will likely get direct support and maybe formally verified. This will be a smaller group as verification is expensive. But would work for things like crypto.
But let’s be clear, the conversation inside companies right now is how to reduce the dependency risk on the thousands of poorly supported projects, and there is growing support to rewrite proprietary alternatives now that it’s cheap to do so.
Ryan Kazanciyan retweeted
Reinsurance expert here: this is a rational decision by Nvidia, not an outlier.
Many of the large tech companies don't do this, other than protection for individual officers.
The size of claims they face and the amount of coverage needed is too large for the market. NVDA, or equivalently, AAPL has more money than the insurer itself, and the premiums charged here would not be worth the self-insurance on average over time.
There's also the invariable coverage disputes in cases like this. Of the 91% who buy insurance, the ones who don't are at the extremes; the very small and the very large.
I know there's a temptation to sensationalize all the headlines right now, but this isn't outlier conduct and it's not going to bankrupt NVDA. The criminal exposure for sanctions is a vastly larger problem than any shareholder lawsuit, and isn't something D&O would cover in many cases anyways.
$NVDA failed to buy D&O liability insurance.
This is just amateur hour for a company of this size.
They are going to be buried under litigation. This just keeps getting worse.
Wiz co-founder and CTO Ami Luttwak says AI can now carry out cyberattacks in 20 minutes that once took skilled attackers weeks or months.
In his view, companies need to continuously attack their own systems with the latest AI models to stay secure. For startups, that means building automated testing and patching into the development process from the beginning.
Ami joins a16z's Sarah Wang and Zane Lackey to discuss what changed during Black October, why patching is falling behind, the cost of continuous security testing, and why better security architecture matters more than simply buying more tools.
0:00 Intro
1:04 Welcome to Black October
3:19 How AI made every day a zero day
6:51 The PHP vulnerability: 70% of the internet
9:19 Why patching can't keep up
11:32 How Wiz uses AI to find vulnerabilities
16:59 The cost of continuous testing
19:23 Cloud security in the AI era
22:18 The red agent: attack yourself first
26:41 Agent swarms attacking your own systems
28:43 What founders must do before they ship
32:15 Where security budgets should go
36:10 Most of what security teams do isn't security
43:34 Scan for Good and public infrastructure
YouTube: youtu.be/PbD9pVXxsYM
@amiluttwak @sarahdingwang @zanelackey
Ryan Kazanciyan retweeted
We’ve published an update on our recent security incident. See a detailed overview of what happened and our remediation steps here: cursor.com/blog/security-inc…
































