- tl;dr sec
- [tl;dr sec] #188 - Security Interview Questions, Secret Scanning Tools, PentestGPT
[tl;dr sec] #188 - Security Interview Questions, Secret Scanning Tools, PentestGPT
Interview questions across a variety of roles, several secret scanning tools, an autonomous pentesting tool using GPT-4
I hope you’ve been doing well!
The “Full Utah” Experience
Last weekend I got to hang out with my friend Scott Piper, and he gave me the “full Utah” experience.
I felt a sense of peace dirt biking down a gravelly dirt road in the desert, with the wind in my hair and a dog running in front of me, chasing a wild antelope.
I waved to the families chilling by their trucks in an outcroppings of trees, who’d smile and wave back.
Later, I saw this on a car:
Not the type of bumper sticker you’d typically see in San Francisco 😆
Which made me think-
Differences, in any relationship (friendship, romantic, political, etc.), can cause friction.
But they’re also a source of growth, and make us better.
I like it.
Scott and I slaying 3 pizzas.
P.S. Thank you so much to everyone who attended Daniel Miessler and I’s webinar yesterday on AI + security 🙏 So many great questions and ideas! I’ll share more next week.
📣 Secure it, torch it, or hope for the best
Material offers a novel approach to classifying and securing sensitive data in employee mailboxes.
We all know that employees’ cloud email accounts are rich and vulnerable targets. We try to implement email retention policies to protect against risk, but employees complain about the impact to productivity.
Material provides the necessary balance: We find and redact sensitive content in emails and bring it back, only when needed, via your pre-existing user-authentication processes (i.e. Okta, Duo, etc).
Customers like Mars, Databricks, and PagerDuty keep sensitive content safe by leveraging Material.
📜 In this newsletter…
AppSec: Modernizing Secrets Scanning, semantic secret scanning tool, tool to extract URLs, paths, secrets from JS
Threat Modeling: CMS Threat Modeling Handbook, Threat Composer
Mobile Security: Visualizing Android code coverage, Semgrep rules for Android security, iOS deep link attacks
Cloud Security: Tool to escalate SSRF in cloud envs, Abusing Overpermissioned AWS Cognito Identity Pools, AWS CloudTrail cheat sheet
Container Security: Kubernetes Security Basics Series
Machine Learning + Security: PentestGPT, NVIDIA’s AI Red Team
Machine Learning: Open source language server for local models, have agents run multiple tools at a time, AI film festival, Midjourney adds “zoom out”, AI-generated websites stealing ad $, GPT-4’s pitch decks outperform humans
Career: Security study plan, security interview questions, awesome AWS security, how James Kettle chooses a research topic, fix the lifestyle you want- then work backwards from there
Misc: Steve Jobs in his own words, Musk/Zuckerberg cage match
Modernizing Secrets Scanning: Part 2–the Semantic Eureka
In Part 1, Avito’s Nikolai Khechumov argues that secrets detection approaches that rely on secrets having known structure/prefix or high entropy are fundamentally limited. In part 2, he describes the value of code-aware tools, which can improve detection rate (200% new findings) and reduce noise. Tool release below.
A tool for secret scanning that uses lexing and parsing techniques to understand code and detect secrets, by Avito’s Nikolai Khechumov. You could also leverage Semgrep rules to do this with no custom tooling effort.
📣 Cloud Detection and Response Survey Report
We surveyed more than 500 security, engineering, and IT practitioners and leaders to understand as much about their cloud environments and security practices as possible. We compared the results with other industries surveys and benchmarks, as well as actual data taken from thousands of customer data published through large cloud service providers.
What did we find? Significant cognitive disconnect in cloud security perception vs. readiness. More than 80% of organizations are confident their current tools and team would cover their organization from a well orchestrated attack, but 95% expressed concern over their ability to detect a threat actor in their environment. Read the survey to learn more.
CMS Threat Modeling Handbook
Overview of the benefits of threat modeling and how it is used to help identify potential weaknesses, by Aquia’s Robert Hurlbut et al. Always interesting to see how big government orgs think about security.
A threat modeling tool to help humans to reduce time-to-value when threat modeling. It provides a prescriptive threat articulation structure, dynamic suggestions, complete threat statement examples, and import/export capabilities to enable persistent storage and sharing.
A collection of Semgrep rules derived from the OWASP Mobile Application Security Testing Guide (MASTG) specifically for Android applications, by Riccardo Cardelli et al.
iOS Deep Link Attacks Part 2 - Exploitation
In part 1, 8ksec discusses the various types of deep link schemas used in iOS apps and how to identify them. In part 2, they delve into exploitation scenarios including phishing attacks, HTML injection, and CSRF vulnerabilities utilizing deep links and inadequate URL validation, as well as mitigation strategies.
Abusing Overpermissioned AWS Cognito Identity Pools
Wes Ladd discusses how identity pools with excessive privileges can allow attackers to authenticate with the AWS Cognito service and carry out restricted actions, including accessing sensitive data, manipulating services, and performing privilege escalation. The article also provides a detailed explanation of the process involved and few commands to exploit this issue.
Kubernetes Security Basics Series: Part II
KSOC continues with their series on Kubernetes Security Basics (Part I - Deployment and container orchestration). This article explores the differences between traditional virtual machines and containers, the anatomy of a container, and security considerations for container and Kubernetes isolation, such as seccomp, AppArmor, SELinux, and third-party open source projects.
Machine Learning + Security
NVIDIA AI Red Team: An Introduction
NVIDIA's Will Pearce and Joseph Lucas present their AI Red Team framework and methodology, which aims to assess machine learning (ML) systems from a security standpoint. The framework brings together offensive security experts and data scientists to detect and address risks in ML systems, while establishing a basis for ongoing enhancement throughout the ML lifecycle.
Multi-Action Agent w/ OpenAI Functions
LangChain updates that now let the LLM select multiple tools to use in parallel (vs only sequentially).
AI Film Festival by RunwayML
Some pretty impressive finalist films.
Midjourney update wows AI artists with camera-like feature
Midjourney’s new v5.2 includes a "zoom out" feature that allows maintaining a central synthesized image while automatically building out a larger scene around it, simulating zooming out with a camera lens.
Junk websites filled with AI-generated text are pulling in money from programmatic ads
>140 major brands are paying for ads on unreliable AI-written sites, 90% were served by Google. NewsGuard is discovering ~25 new AI-generated sites each week and found 217 in 13 languages since April.
GPT-4 Outperforms Humans in Pitch Deck Effectiveness Among Investors and Business Owners
Pitch decks were 2x more convincing,
Overall, investors and business owners say GPT-4 generated pitch decks are 2x more convincing than those made by humans.
Overall, investors and business owners were 3x more likely to invest after reading a GPT-4 pitch deck than after reading a human one.
1 in 5 investors and business owners pitched by GPT-4 would invest $10,000 or more.
Considering the required time investment
Exploring topics that push you beyond your comfort zone
Focusing on novel attack techniques
Consider the applicable audience- building on a large body of recent research may make it hard for many to follow
Prioritize personal development
Check out James’ upcoming talk at Black Hat USA and DEF CON 31 "Smashing the State Machine: The True Potential of Web Race Conditions."
The Most Important Piece of Career Advice You Probably Never Heard
Cal Newport is the man. His book So Good They Can’t Ignore You is one of the best books on career/life I’ve ever read, and I’ve recommended it to dozens of people (video summary by Ali Abdaal).
My ranking of every Shakespeare play - Chris Wellon’s thoughts after seeing all 37 plays live. Includes some watch links. Love it.
arwes - A Futuristic Sci-Fi UI Web Framework.
Make Something Wonderful: Steve Jobs in his own words
A curated collection of Steve’s speeches, interviews and correspondence. Steve shares his perspective on his childhood, on launching and being pushed out of Apple, on his time with Pixar and NeXT, and on his ultimate return to the company that started it all.
Elon Musk Proposes ‘Cage Match’ With Mark Zuckerberg
Elon: “I’m up for a cage match if he is.” Zuckerberg responded with a screenshot of Mr. Musk’s tweet and the caption, “Send Me Location.” What a world we live in.
✉️ Wrapping Up
Have questions, comments, or feedback? Just reply directly, I’d love to hear from you.
If you find this newsletter useful and know other people who would too, I'd really appreciate if you'd forward it to them 🙏
Thanks for reading!