Machine Learning talks in RSA Con 2017

Machine Learning talks in RSA Con 2017

The RSA Conference is one of the most widely attended security conferences in the world, and the 2017 edition, held in SFO, concluded just about 10 days ago.

There were close to 20 presentations this time, around using Machine Learning (referred to as ML hereon in this post) in detecting/preventing cyber attacks of various kinds. And in this post I share my take and a summary (detailed in some cases) on the Top 10 talks on ML.

Some of these talks, especially research projects, require a detailed discussion and analysis, but I’ve tried to do justice to them by keeping my summary as detailed as possible. I plan to dive deeper into some of these topics, in the future.

Note: I have included a link to the original Talk (presentation or video) wherever I could find them, so do check them out.

  1. A Vision for Shared, Central Intelligence to Ebb a Growing Flood of Alerts

Dan Plastina, who heads Threat Protection at Microsoft, gave a talk on striking a balance between using ML in threat detection and also in Incident Management/Orchestration process, using linked Graph and chat Bots, in “SecOPS Console”, to better manage the growing flood security alerts. What I found interesting in this talk is the mention of a whole gamut of Microsoft products, many of which are familiar to us, like AD, Office, Azure security center. But I couldn’t find if Dan was also referring to an IR Orchestration tool that Microsoft has built or is int roadmap. Also, I see that R is being tightly integrated into various Microsoft products.

An interesting talk indeed, and here is the link to the original talk.

2. Advances in Cloud-Scale Machine Learning for Cyber-Defense

Another talk from Microsoft; this one by Mark Russinovich, the CTO for Microsoft Azure. This one was quite a deep dive into how Microsoft uses ML in detecting cyber attacks on the Azure platform. My quick notes below:

  • He started off with some metrics:
    • More than 10,000 location-detected attacks (detected/reflected attacks) – I didn’t understand what exactly he meant here.
    • 1.5 mil compromise attempts deflected
  • Red team and Blue team kill chain – it was interesting to see how each of the blue team’s “response” are mapped to read team’s malicious action stages
    • Attack disruption shows execute stage before move stage
  • Their “supervised” learning approach enables detection with minimal FP – this is an interesting claim
  • “Attack disruption” requires us to think of ML beyond detection
  • He also covered properties of successful ML solution – adaptable, explainable, actionable, results in successful detection
  • Framework for a successful detection – honestly this is one of the best and simple visual representation/explanation of how an ML based solution should look like. He also talks about two Case studies where IPFIX data is used as a training set, and detecting malware using a combination of Rules and ML
  • Then he goes deep into Case study 2 where he talks about the algorithms and compares fingerprint based detection to behaviour based.
  • Triage incidents not alerts – very valid point
  • In a nutshell – attack disruption means to shorten blue team kill chain

The Video to the original talk is available here.

3. Combatting Advanced Cybersecurity Threats with AI and Machine Learning 

This one was by Andrew B. Gardner, Head of Symantec’s ML Program. My notes below:

  • Interesting perspective shared here, but a bit high level.
  • He starts off with comparing AI & ML and how they differ in cyber – interesting point about the use of ML in cybersecurity, rather than AI, for various reasons:
    • complex sequential data
    • not human intuitive (logs)
    • labels are expensive (scarce)
    • closed research models
  • Typical use of ML in cyber today: collect data sets > training algorithms > build a model > updated classifiers > ingested to another “threat detector”
  • Though the advantages of using ML in cybersecurity are good, Andrew poses interesting argument around what are disadvantages of using ML in cyber security:
    • dependency on data (quality, completeness), and system
    • adversaries also have access to ML
  • ML at Symantec
    • some interesting approaches shown, about optimizing models – True positive to false positive ratios (ROC) and how to optimize them
    • use of string scoring services – Charlatan

Link to the original talk is here.

4. Automated prevention of ransomware with Machine learning and GPOs

This talk was by Rod Soto (Security Researcher at Splunk) and Joseph Zadeh (Security Data Scientist at Splunk). My notes below:

  • Rod and Joseph started with some key aspects of detecting ransomeware in the “new age” – behavioural modeling, unsupervised ML, anomaly detection and leveraging big data
  • Use of Aktaion tool kit for building the detection system
    • Take PCAPs of known (labeled) exploits and known (labeled) benign behavior and convert them to bro format
    • Convert each Bro log to a sequence of micro behaviors (machine learning input)
    • Compare the sequence of micro behaviors to a set of known benign/malicious samples using a Random Forest Classifier
    • Derive a list of indicators from any log predicted as malicious
    • Pass the list of IOCs (JSON) to a GPO generation script
  • Key is to focus on delivery of exploit (in addition to using system specific and call back specific behaviours) – following key steps were covered:
    • training a model (Random forest algorithm used in this case), to detect exploit delivery, using known malicious indicators
    • tuning the hyper parameters – risk factor, age, session time, entropy, etc.
    • model classifier built with 6 trees
    • the model will start generating output that separates signal from noise (they use the Splunk MLTK in this case)
    • link it to GPO scripts to automate the response procedures via power shell (active defense)
  • Training set and test data used in the demo include datasets from Contagio, DeepEnd Research, Ransomware samples with some call back and file system level indicators, labelled benign http user traffic (anonymized bluecoat logs)
  • The talk then ends with a PoC demo of this whole workflow
  • Summary: ML + GPO = Active Defense

Link to the original talk here.

5. Big Metadata: Machine Learning on Encrypted Communications

This one was by Jennifer Fernick and Mark Crowley, Security Researchers from University of Waterloo. My notes below:

  • This is derived from a research project, and was a very interesting session where not just the application of ML in cybersecurity was discussed, but also the inverse – security in the computational functions of ML
  • In this talk Jennifer and Mark talk about
    • ML research in cyber security – applying ML to problems in cybersecurity
      • using ML in cyber security
      • cybersecurity for ML – adversarial ML – study of ML systems in adversarial environments, where an attacker might train the system in hopes of modifying its behaviour to allow for an attack
      • a mid way – secure ways of computing ML functions
    • Candidate problems depend on information sources
    • Metadata – how can we use metadata for building the training set, while keeping privacy concerns intact?
    • ML 101 – a crash course
    • Their work in the field, and
    • Future direction
  • In the “security for ML” topic, there were some very interesting concepts presented – secure multi-party computation, privacy preserving data mining, homomorphic encryption, differential privacy. All these are deep mathematical and computation fields in themselves and definitively requires intensive reading. And so I am going to stop at that!
  • In the “ML in cybersecurity” topic, some fundamental questions were called out – what problem am I trying to solve
    • securing my learning data?
    • learning my security data?
  • On “ML 101” topic, they give an excellent crash course on ML and how to use it in cybersecurity
    • use of clustering (unsupervised learning) and classification (supervised learning)
    • system design and algorithm choices
  • Their work in ML – use of ML on encrypted data – analysing private and public communication networks to detect anomalies
  • I have to confess I found this talk to be the most difficult to thoroughly grasp, as the talk was research oriented and definitely calls for an in depth reading on each of the sub-topics covered. A great presentation indeed!

Link to the original talk here.

6. Applied Cognitive Security: Complementing the Security Analyst

This one was by Vijay Dheap, Program Director, Cognitive Security at IBM.

  • This talk was primarily about IBM’s Cognitive security product built on Watson their Qradar Security intelligence platform, and how it can help a Security Analyst better detect, analyse and respond faster to security incidents.
  • The presentation was high level and didn’t get into the details of how Cognitive Security with IBM Watson actually works. For ex., what algorithms are used, and what are the typical hyper parameters, and how they are used in conjunction with contextual feeds (vulnerability, asset, identity, behaviour) to detect security incident more effectively.
  • The presentation did cover one case study with a Botnet use case, but didn’t reveal much information on the inner workings (atleast some indication) of how ML and Watson’s AI detected this incident.
  • A good “high level” talk over all.

Link to the original talk here.

7. Dealing with Millions of Anomalies

This one was by Chris Larsen, Threat Researcher with Symantec

  • The talk was about detecting malicious traffic, by using ML (anomaly detection), and TI data
  • He first approach to handle the issue of picking “interesting anomalies” in millions of anomalies, is to pick “One Hit Wonders” and “One Day Wonders”, and then investigating them further by using various attributes (IP address licenses, ports used, are they DGA, etc.)
  • Once we have this “interesting anomalies” filtered out, then run it against good TI, to pick the most probable malicious traffic.
  • Summary: good TI is the key, and a good place to start, are TI that has malware/attack “families” context, industry/vertical/geo context.
  • Definitely an interesting talk with real world examples like using IOC data for Angler and Magnitude exploit kits, to filter out “most probable” malicious traffic, and then drilling further down from there.

There is a video of Chris’s gal available here. Definitely worth watching.

8. Machine Learning: Cybersecurity Boon or Boondoggle

This one was by Dr. Zulfikar Ramzan, CTO of RSA.

  • The talk starts at an elementary level, covering the fundamentals of ML and its use in Cyber security.
  • But towards the end, Zulfikar covered some very interesting facts/tips/best practices while using ML in cyber security. For ex.:
    • The importance of ROC (Receiver Operating Characteristic Curve) while making a trade-off between True positive and false positive classifications.
    • ML (in this case unsupervised) only is helpful in detecting bad “actions”, and not bad “intent”, and thus resulting in calling out lot of legitimate “unusual actions” as “bad/malicious”.

Link to the original talk here.

9. Applied Machine Learning: Defeating Modern Malicious Documents

This one was by Evan Gaustad, Sr. Manager, CSIRT – Target.

  • The talk basically starts with typical vulnerabilities exploited in Microsoft Office (Macros), and some examples of the attack lifecycle using malicious documents itself
  • Evan then gets into the details of the project he has been working on, where he used supervised ML (classification) to detect malicious documents. There is a video recording of his talk here, and I strongly recommend it. He covers a lot of details of how the model and its classifier actually works, with examples.

There is a video of Evan’s talk available here. Its a must watch.

10. An Introduction to Graph Theory for Security People Who Can’t Math Good

This one was by Andrew Hay, CISO, Data Gravity.

  • Though this talk didn’t actually cover how ML is used in detecting/preventing cyber attacks, it was a great crash course on Graphs theory (for the non-mathematicians amongst us), and how it can be extremely useful in visualising an attack lifecycle
  • Application of Graphs in security context
    • incident response – use of Google’s Fusion tables to visually represent the communication/interactions between user and entity in a security incident
    • actor tracking – detecting the source of a phishing campaign – using the IOCs available, use Maltego (CE)
  • What was interesting in this talk was – it is so easy to build a visual representation of the interaction. However, it can get way too complicated to interpret, due to a bad choice of dataset and the “vertices” (nodes) and “edges” (connections) in it.

The link to the original talk is available here.

 

Thanks for reading through my point of view RSA Con USA 2017. I hope I was able to provide byte sized (mega!) summary of some of the most interesting talks in this conference this year.

PS: Do subscribe to this blog, to get notified the moment I publish my next post.

On Apple’s Bug bounty program

On Apple’s Bug bounty program

The Head of Security Engineering and Architecture at Apple, Ivan Krstić, announced to Black Hat attendees last week, that Apple will begin offering cash bounties of up to $200,000 to researchers who discover vulnerabilities in its products.

Krstić’s talk at Black Hat was definitely interesting and covered a good breadth of the technical measures that Apple has been taking in making iOS secure, from grounds up. The presentation also included a level of technical detail and disclosure of security—here, related to AutoUnlock, HomeKit, and iCloud Keychain—that has been mostly absent in the past at conferences, according to those present.

Apple being so open and forthcoming, about their security architecture, is somewhat unusual, but definitely welcoming.

Now, about the the bounty program itself, it will initially be limited to about two dozen researchers who Apple will invite to help discover difficult-to-uncover security bugs in five specific categories:

Screen Shot 2016-08-24 at 9.18.30 PM

Each of these aspects represent key threat vectors for attacks by governments and criminals alike. While iOS has never had exploits spread significantly in the wild, jailbreaking the software has made use of various methods of running arbitrary code in iOS. In another Black Hat presentation, the makers of the Pangu jailbreak for iOS 9 (fixed in 9.2), described how they achieved that kind of code execution.

Until now, there’s been no known extraction of data from Secure Enclave, the dedicated hardware in iOS devices with an A7 or newer processor that acts as a one-way valve to store fingerprint characteristics and certain data associated with Apple Pay. It is also used to prevent downgrading iOS to exploit a bug in a previous release. iCloud, which has been in the media sometimes for the wrong reasons, have had some accounts compromised in the past through certain weak password entry endpoints and social engineering of celebrity accounts, there has been no reported breach of iCloud servers itself.

Going by these clearly laid out vulnerability categories and qualification parameters, I see that Apple’s program sets clear objectives – find exploitable bugs in key areas. It makes complete sense, because proving exploitability with a repeatable proof of concept, takes lot more effort than merely finding a vulnerability. If the bug is found to have significant impact on security, then Apple will pay the researchers a fair value for their work. By doing this, Apple aims to learn how to improve a bug bounty program, over a period of time, and derive maximum value out of it.

The end result is – high-quality vulnerabilities (and their respective exploits) discovered, by researchers and developers who Apple believes have the skills and the right intentions to help advance product security. Bounty fees at other companies range from a starting point from $100 to $500, and are capped at from $20,000 at Google to $100,000 at Microsoft, clearly indicating the focus being quantity, unlike Apple’s focus on quality and difficult to discover, exploit and reproducible vulnerabilities.

Many major tech companies, like Google, Facebook, Microsoft, Adobe, and SAP, have been running Bounty programs for years. But there is a reason for Apple not getting into the Bounty business until now, even if security has always been a priority for them and iOS is way more secure, grounds up, than other competing mobile OS platforms today. That reason is primarily to ward off governments and underground hackers who merely want to make money, by not being in a position to negotiate with them. The disclosure by the United States government on last week that an unknown third party had approached it — and not Apple — to help open a controversial iPhone only highlights how the giant company approaches bug-hunting efforts and security differently from the rest of the tech industry.

Asked by the audience at Black Hat why Apple waited so long to launch a bounty program, Krstić said the company has heard from researchers that finding critical vulnerabilities is increasingly difficult, and it wanted to reward those who take the time to do it.

I have been following Apple closely since 2009, when I bought my first Apple product – an iPhone 4 (the last phone Steve Jobs personally launched). Being a Security Consultant myself, I have always wondered to how Apple builds their software to be far more secure than other operating system platforms. And this has been true from the very beginning of Mac (built on a strong Unix base), And so I have always tried to understand iOS and Mac security a bit deeper, but Apple has always been secretive about sharing information, just the way they are about their product strategy and roadmap. So this new development with the Bounty program and the overall incharge for Product security at Apple making a presentation at Blackhat, is very exciting to me.

I am looking forward to understanding how Operating System security is best handled, from a company that makes the best software and hardware in the world today.

Notes:

  1. Krstić’s presentation at Black Hat is available here
  2. The video of the talk has been published recently on YouTube

 

Feature Image courtesy: blackhat.com

Need a security expert? You got to hire a coder!

Need a security expert? You got to hire a coder!

As security (cyber) becomes more and more important, to businesses, governments, and also to our personal lives, the need for good security engineers and researchers is increasing at a rapid pace.

This is true whether one is working in an entry-level position or is already a senior researcher.

It is often said in the security industry that “It is easier to teach a developer about security than it is to teach a security researcher about development (coding).”

Information security professionals are used to seeing, experiencing and talking about failures in the industry. This usually leads them to assume that badly written (vulnerable) code is always the product of unskilled developers. If these professionals have never been exposed to software development, even at a small scale, then they do not have a fair understanding of the complex challenges that developers face in secure code development. And I think that a security professional cannot be effective in designing detective and preventive security controls (tools, architectures, processes) if he or she doesn’t appreciate these challenges.

Let me illustrate that with an example- ‘code injection” attacks against NoSQL databases versus SQL databases. Simply put, SQL and NoSQL databases both collect, organize and accept queries for information, and so both are exposed to malicious code injections. So, when NoSQL databases became popular, people were quick to predict that NoSQL injection would become as common as SQL injection. Though that is theoretically true, developers know that it’s not that simple.

If you take sometime out understanding NoSQL databases, you will quickly realize that there are a wide variety of query formats, from SQL like queries (Cassandra), to JSON based queries (MongoDB, DynamoDB), and to assembly like queries (Redis). And so security recommendations and tools for a NoSQL environment have to be targeted to the individual server that is underneath. Also, your security testing tools must have the injection attacks that are in the format of that specific database. And so one cannot blindly recommend controls or preventive measures, without understanding that the vulnerabilities are not available on all platforms. Encoding recommendations for data will be specific to the database type as well. This OWASP article explains how one can test for noSQL injection vulnerabilities.

This is all the knowledge that one can learn by digging deep into a subject and experimenting with technologies at a developer level. And so people with development backgrounds can also, often times, give better technical advice.

If one looks at the people leading security programs or initiatives at companies like Apple, Facebook, Google, and other large successful tech companies, many of them are respected because they are also keeping their hands on the keyboards and are speaking from direct knowledge. They not only provide advice and research but also tools and techniques to empower others in the same industry.

So to summarise, I would like to say that whether one is a newly graduated engineer or a senior security professional or a security researcher, one should never lose sight of the code, as that is where it all begins!

 

 

Picture courtesy: http://www.icd10forpt.com

Verizon’s acquisition of Yahoo

Verizon’s acquisition of Yahoo

TechCrunch just reported that Verizon has acquired Yahoo for $4.83 billion. 

This definitely is a shocker and I am sure many would agree with me. Not many of us were expecting Marrisa Mayer to call it a day by dropping the ball so soon. 

One of the most important companies of the first dot-com boom, Yahoo, has reached the end of its life as an independent company. This deal represents a stunnin decline for a company that was valued at more than $100 billion at its its peak in 2000. 

Marissa’s roots as an engineer at Google, definitely helped in improving the brand value with software programmers and technology users alike, and she did make an effort to beef up Yahoo’s technical talent. She instituted a regorous recruitment process and it worked hard at hiring computer scientists from some of the best universities. But there is little sign that these moves changed the culture at Yahoo or improved morale among the programmers working there. They always saw and projected themselves as a “media company” and not a “technology company”. I am not sure if it played out well for them, as its attempt to be a tech company and a media company at the same time, resulted in an organisation that was less than the sum of its parts. 

I strongly believe that one reason why Verizon was a strong contender was that they have done this before; Verizon acquired another struggling Internet company last year. Like AOL, Yahoo makes a lot of money by creating Internet  content and selling ads against it. So from Verizon’s perspective, this definitely looks like a logical step.

With respect to Mayer’s future at Yahoo, I am sure she is pursuing opportunities outside, as the statement that Yahoo released about this deal, “Yahoo will be integrated with AOL under Marni Walden, EVP and President of the Product Innovation and New Businesses organisation at Verizon”, makes it evident that Marissa Mayer’s future lies outside of Yahoo. 

I wish her all the best, and am sure she will build something very interesting soon in the tech business.



Picture courtesy: TechCrunch.com

Cyber weapons and Nuclear weapons

A good essay pointing out the weird similarities between cyber weapons and nuclear weapons. 

On the surface, the analogy is compelling. Like nuclear weapons, the most powerful cyberweapons — malware capable of permanently damaging critical infrastructure and other key assets of society — are potentially catastrophically destructive, have short delivery times across vast distances, and are nearly impossible to defend against. Moreover, only the most technically competent of states appear capable of wielding cyberweapons to strategic effect right now, creating the temporary illusion of an exclusive cyber club. To some leaders who matured during the nuclear age, these tempting similarities and the pressing nature of the strategic cyberthreat provide firm justification to use nuclear deterrence strategies in cyberspace. Indeed, Cold War-style cyberdeterrence is one of the foundational cornerstones of the 2015 U.S. Department of Defense Cyber Strategy.

However, dive a little deeper and the analogy becomes decidedly less convincing. At the present time, strategic cyberweapons simply do not share the three main deterrent characteristics of nuclear weapons: the sheer destructiveness of a single weapon, the assuredness of that destruction, and a broad debate over the use of such weapons.

Questions to ask before you get your first Threat Intel data source

Anton Chuvakin (one of the leading Gartner experts in the Threat Detection space) had a recent blog post on some of the key questions one must ask while identifying the first threat Intel data source. 

Here is the list

  • What is the my primary motivation for getting TI, such as better threat detection, improved alert triage or IR support?
  • Where do I get my first threat intel source [likely, a network indicator feed, IP/DNS/URL]?
  • How do I pick the best one(s) for me?
  • Where do I put it, into what tool?
  • How do I actually make sure it will be useful in that tool?
  • What has to happen with the intelligence data in that tool, what correlation and analysis?
  • What specifically do I match TI against, which logs, traffic, alerts?
  • What you have to do with the results of such matching? Who will see them? How fast?
  • How to I assure that the results of matching are legitimate and useful?
  • What do I do with false or non-actionable matches?
  • How do I use intel to validate alerts producted by other tools?
  • Do I match TI to only current data or also to past log/traffic data? How far in the past do I go?

The post is worth a read, as he has linked his earlier posts on this topic in this blog post. Do note that the white papers he has has linked requires GTP access. 

A great list of curated Threat Intel resources

A great list of curated Threat Intel resources

I recently found this Github Repo, put together by Herman Slatman, which consists of a list of very useful and curated Threat Intelligence resources.

The list is broken down into following five categories:

  • Sources
  • Formats
  • Frameworks
  • Tools
  • Research, Standards & Books

This is a great resource for anybody starting to dwell into the Threat Intelligence discovery, consumption and classification, as it is an ocean out there, and a lot of these “Indicators” can be noise.

 

Picture Courtesy: depositphotos.com

Interesting Data Science projects of 2015

Interesting Data Science projects of 2015

Here is a list of some really interesting Data Science projects of 2015. Thanks to Jeff Leek from @simplystatistics for putting this together. 
Some of my picks from the list are:

* I’m excited about the new R Consortiumand the idea of having more organizations that support folks in the R community.

* Emma Pierson’s blog and writeups in various national level news outlets continue to impress. I thought this oneon changing the incentives for sexual assault surveys was particularly interesting/good.

* As usual Philip Guo was producing gold over on his blog. I appreciate this piece on twelve tips for data driven research.

* I am really excited about the new field of adaptive data analysis. Basically understanding how we can let people be “real data analysts” and still get reasonable estimates at the end of the day. This paper from Cynthia Dwork and co was one of the initial salvos that came out this year.

* Karl Broman’s post on why reproducibility is hard is a great introduction to the real issues in making data analyses reproducible.

* Datacamp incorporated Python into their platform. The idea of interactive education for R/Python/Data Science is a very cool one and has tons of potential.

Picture Courtesy: kdnuggets.com

Adopting OODA Loop in Intrusion Detection & Response – it’s more than speed

Adopting OODA Loop in Intrusion Detection & Response – it’s more than speed

Here is a great post by Richard able, on the concept of using OODA loop in Intrusion Detection and Response.

I have included some interesting lines here:

It is not absolute speed that counts; it is the relative tempo or a variety in rhythm that counts. Changing OODA speed becomes part of denying a pattern to be recognized…

The way to play the game of interaction and isolation is [for our side] to spontaneously generate new mental images that match up with an unfolding world of uncertainty and change…