CISO MindMap 2025: What do InfoSec Professionals Really Do?

Job of a Chief Information Security Officer (CISO) is complex. Many individuals outside the realm of cybersecurity often underestimate the intricacies involved in a security professional’s role. Since its inception in 2012, the CISO MindMap has served as a valuable educational resource, offering insights into CISO responsibilities and aiding security professionals in crafting and enhancing their security programs. Continuously adapting to reflect the evolving landscape of cybersecurity, the CISO MindMap has been updated to accommodate the latest developments in the field. Here is the most recent iteration of the CISO MindMap for 2025, featuring numerous enhancements and fresh recommendations for the next 12-18 months covering the year 2025-26.

Don’t forget to review recommendations for 2025-26 described below and to subscribe to my blog.

Download High Resolution PDF for Printing

Click here to download high resolution PDF version of the CISO MindMap 2025

Summary of Changes

With accelerating digitization of businesses, the responsibilities of security professionals are only increasing. Technology is changing fast, bringing new ways of doing business, continuous adoption of Cloud, and extremely fast evolving AI/GenAI technology. Not only the Infosec professionals are “expected” to deeply understand these technologies, they are also tasked with providing policies/guidance on how to secure them. For this reason, every year you find new items on the CISO MindMap. At the same time, some items are updated or removed from the CISO MindMap depending upon their relevance or obsolescence. In the latest CISO MindMap, modified and new items are marked in red color for your convenience.

Expiration Date – A common issue is that many professionals still have older CISO MindMap copies. Like last year, I added an “expiration date” to let people know when they should stop using a particular version. The expiration date for the 2025 CISO MindMap is the end of September 2026. The next version will be published before the current version expires.

CISO MindMap Update Methodology

Every update to the CISO MindMap undergoes thorough consideration, research, and attention to detail. In addition to my ongoing engagements with industry leaders, various methods are employed to ensure we capture the pulse of the cybersecurity landscape:

  • Conducting interviews with experts.
  • Distributing surveys to gather insights.
  • Leveraging LinkedIn for targeted questions and discussions.
  • Analyzing feedback and comments from previous versions of the CISO MindMap.
  • Staying abreast of industry news and conducting in-depth analysis.
  • Review of research reports published by reliable organizations.

Furthermore, I’d like to express my gratitude to the contributors whose valuable insights have enriched this endeavor. The ‘Acknowledgments‘ section of this blog post includes their names and LinkedIn profiles as a token of appreciation.

Are you Accountable for Everything included in the CISO MindMap?

Security is inherently a collaborative effort, and the role of the CISO entails providing consultative guidance in various areas outlined within the CISO MindMap. It’s crucial to discern between areas where direct ownership and accountability are necessary and those where consultation is the primary function. Within any organization, numerous stakeholders are involved, and a common pitfall is the lack of clearly defined boundaries for each role.

My recommendation is to establish a RACI (Responsible, Accountable, Consulted, Informed) matrix, which serves as a standardized methodology for delineating roles and responsibilities among stakeholders. By mapping out tasks and corresponding roles within this framework, clarity is achieved, ensuring that each stakeholder understands their level of involvement and contribution to the overall security landscape.

So yes, a CISO is probably not the only role to cover everything in this MindMap, hence, building alliances is important for success.

Focus Areas and Recommendations for 2025-2026

Each year, I offer my recommendations as a practitioner, drawing insights from discussions with information security leaders. My approach strives for objectivity, steering clear of hype and focusing solely on data-driven research. Though unintended biases may exist, the goal remains to propose actionable steps viable within a short to mid-range timeframe. These recommendations do not constitute future predictions; rather, they address the immediate needs for enhancing security programs.

Selecting a concise set of recommendations is always a challenge for me. While the list provided below may be longer than I’d prefer, I aim to offer a comprehensive array of considerations for your reflection. Followers of the MindMap will notice both recurring suggestions from previous years and new additions, providing a blend of continuity and fresh insights.

I’m genuinely interested in hearing your perspective on these recommendations and understanding whether they resonate with your experiences and insights. Your feedback, whether in agreement or disagreement, provides valuable insights that can help refine and improve our approach to addressing security challenges. So, please feel free to share your thoughts and insights on these recommendations—whether you support them or have reservations—so we can engage in a constructive dialogue to further enhance our security strategies.

Recommendation 1: It is Time for Securing GenAI

A lot has happened since last year when I made a recommendation to “adopt a cautious approach towards GenAI”. We have seen many GenAI players releasing more advanced LLMs while new players enter the market. Now LLMs can do better reasoning, are faster, and need less time/resources to train and operate. New entrants such as Deepseek have made ripples causing swings in stock markets. The new hype seems to have shifted from chatbots to AI agents. GenAI is proving to be useful in areas such as code assistants/code generation, marketing and publications. However, my own research indicates that people are still not very comfortable letting AI agents make autonomous crucial decisions without human oversight. Use of GenAI as assistance is getting traction, but autonomy still seems far in future. At the same time, governments around the globe are setting up standards and regulations for responsible use of AI. 

On the security front, organizations like OWASP, NIST and others have published recommendations for protecting AI from attacks such as supply chain, prompt injection, data poisoning and others [1].

What does this mean for CISO and security leaders? Following are some recommended actions.

  • Setup standards and governance for GenAI.
  • Research, validate, and approve LLMs (and other AI models) for use within organizations.
  • Define what it means to have a responsible use of AI within an organization.
  • Create a process to review use cases to ensure standards are adhered to.
  • Update vulnerability management and AI application penetration testing processes to cater new needs for GenAI.
  • Train security professionals on how GenAI, LLMs, RAG and Agents actually work.
  • Establish a training and awareness program for IT staff.
  • Explore open source options for GenAI (which are already providing viable solutions).

I have published a few blog posts about understanding GenAI threat and risk categories [2]. In addition to securing GenAI, it is also a good time to start looking into how to take advantage of GenAI in Cybersecurity operations through plausible use cases [3].

Recommendation 2: Consolidate and Rationalize Security Tools

This is a continuation of a recommendation made last year as the situation in this particular area has not changed much. An average organization is using a large number of security tools, by some estimates as high as 47, according to a Ponemon survey. Yet many security leaders don’t know if Cybersecurity tools are working. Accumulating more security tools doesn’t necessarily lower risk; rather, it amplifies the necessity for maintaining expertise within security teams. Misconfigured security tools may become a liability in themselves. There is a need for consolidation and rationalization of security tools by deeply exploring Return on Investment (ROI) of these tools. You may be surprised by the amount of shelfware within security programs. In some cases open source tools may work just fine as well and could be a good replacement for commercial tools.

When it comes to security tools, many organizations are adopting one of the two most popular strategies.

  1. Ecosystem approach – A single vendor with an ecosystem of tightly integrated security tools. This approach provides, at least in theory, a single view of risk. Vendor lock-in is a major concern, however. There is also a risk associated with security breaches and vulnerabilities that could impact the whole ecosystem.
  2. Best of the breed approach – In this approach an organization uses the best available security tool for a certain purpose. This approach brings its own pitfalls of creating silos and problems in building an organization wide view of risk.

If you go with the “best of the breed approach”, my strong recommendation is to opt for the tools that have strong API capabilities so that you can build an internal “glue application” to integrate data available from these tools.

The guiding principle is that no tool should be worth more than the value they bring and the risk they reduce.

Recommendation 3: Identify and Manage Security Debt

There is no secret that businesses are becoming more digital, new software applications are being built and are becoming more interconnected and complex. Security teams find a higher rate of software flaws, infrastructure vulnerabilities, and architectural issues than they are able to fix. At the same time, development teams are under pressure to churn out new features and new applications with less attention on fixing old and existing issues. The result: organizations are building security debt at a faster pace than ever.

My recommendation for CISOs and other security leaders is to initiate a program to identify and measure the security debt, and estimate time/money it will take to close it, or at least not pile up more. I would also suggest making the security debt (in terms of dollars) part or a reporting metric for boards to raise awareness.

Recommendation 4: Ransomware and Cyber Resilience

Ransomware continues to be a major problem for security professionals and, at the same time, a very profitable business for bad actors. This problem is especially huge for industries that provide crucial human services such as hospitals, and for critical infrastructure organizations. The mindset of “it is other people’s problem” still exists and much more is needed to prepare for effectively responding to ransomware attacks. Thankfully, the awareness has increased over time, which is a progress in itself.

To prepare for and respond to ransomware attacks, here are some very foundational recommendations. These recommendations are just a starting point and a continuous review of defences is needed to properly safeguard against ransomware attacks.

  • Conduct a business impact analysis.
  • Conduct exercises to thoroughly test backups that can be restored in a reasonable time. Note that backups could also be attacked/ ransomed.
  • Purchase cyber insurance and make sure ransomware is included in the coverage.
  • Focus on resilience and ability to restore operations. Take a systems thinking approach.

Recommendation 5: Create Meaningful Metrics

The CISO MindMap includes areas such as metrics and reporting. Traditionally a major focus has been on operational metrics such as number of system patched, number of vulnerabilities fixed, number of events blocked, etc. The recommendation for this year is to turn this focus on more meaningful metrics that actually show progress towards managing risk to an appropriate  and acceptable level.

While the concept of “meaningful” may slightly vary from organization to organizations, our friend Wade Baker has made few suggestions about meaningful metrics like the following:

  • Control effectiveness
  • Risk reduction
  • Overall program performance

I will also recommend that automation of a metric is as crucial as the metric itself.

Recommendation 6: Improve Cyber Hygiene

While many are busy in GenAI related initiatives, we must not forget the basics. Many security professionals are painfully aware of the fact that, more often than not, security incidents and data breaches are a result of missing basic cyber hygiene. Please continue work on these basics such as:

  • Enhance visibility into digital footprint with a target of obtaining full visibility.
  • Shrink the attack surface for your organization as much as possible.
  • Enhance API security. APIs are fueling growth of complex systems and are becoming an even bigger part of the overall infrastructure.
  • Manage and reduce complexity. Remember complexity is the enemy of security.
  • As an industry we are not very good at Third Party risk management. Keep finding better ways of doing so.

Others Recommendations

While the above six are the “official” recommendations for CISO MindMap 2025, I just want to acknowledge suggestions, insights and initiatives from the contributors as listed below.

  • Privacy by design.
  • Understanding and dealing with AI powered threats.
  • Quantum strategy and planning.
  • Demonstrating and assessing assurance over AI models; upskilling the security team to be successful in the age of AI; building business-centric processes and operating models to deliver business value.
  • Proactively assess the value of each control and develop plans to sunset at least 5% of controls within 24 months.
  • Current approach to third party risk management delivers very little value. Automation in Third Party Risk Assessments and explore use of AI for assistance.
  • Changing security culture towards a service oriented approach. “Shake up the game” and impact behavior change.
  • Effective board presentations.
  • A systems approach to Cyber risk taking into account geopolitical challenges, economy, potential recession, and layoffs.
  • Learning from Cyber incidents.
  • A realistic, balanced, and practical approach towards DLP.
  • Real time pro active cyber risk monitoring via singular CISO Dashboard (Like Salesforce for CFOs and Hubspot for CMOs).

How to Use CISO MindMap?

Have you ever been asked to explain what you do as a security professional? The CISO MindMap originates as an effort to answer this question and has evolved since then. It is a good tool for addressing this question and clarifying the intricacies of the role. Many professionals have attested to its efficacy in elucidating the complexities of the CISO position, particularly when communicating with a business audience.

Here are some ways in which the CISO MindMap proves to be immensely valuable:

  • Facilitating conversations with fellow technology professionals.
  • Instrumental in the design and refinement of security programs.
  • Adopted by certain security vendors for raising awareness.
  • Employed in CISO group discussions and community meetings.
  • Aiding aspiring security professionals in understanding the industry landscape and charting their career paths.
  • Serving as an educational and awareness-raising tool.

Obviously there is a lot on this MindMap. The stress on people who have these responsibilities is real. If nothing else, this MindMap should help leaders recognize that stress and do something about it. I covered this topic (stress) in my latest book Cybersecurity Arm Wrestling: Winning the perpetual fight against crime by building a modern Security Operations Center (SOC) as well.

What are They Saying?

Following is a sample of comments on CISO MindMap post from 2024.

  • Jack Jones wrote – “Really great piece of work, Rafeeq!
  • Jerich Beason CISO wrote – “This mind map has helped many CISOs explain the depth and breadth of their roles to so many executives…Thank you for allowing me to play a small role in this latest evolution.
  • Gary Hayslip CISO Softbank wrote“thank you, my friend, for giving me the opportunity to participate in this year’s mindmap; I really like how the GenAI section came together. As always, thank you for continuing to develop this amazing resource and providing it to our #community”
  • Omar Khawaja CISO wrote“What a comprehensive list!”
  • Atif Yusuf wrote“Great job. Once again ! I especially like inclusion of “Business Resilience” and “Building a brand“
  • Ross McNaughton CISO at Gulf Bank wrote – “As always, a great resource showcasing our field. Appreciated for all your efforts in maintaining and enhancing it constantly

Following are some comments on CISO MindMap LinkedIn post from 2023.

  • Michael Restivo“One of the most valuable documents around. Love this Rafeeq.”
  • Chris Novak“Always a great asset! Thanks for all that you do for the cybersecurity community!”
  • Muath AlHomoud“Great insight Rafeeq Rehman always inspiring”
  • Christophe Foulon “Insightful as always, Rafeeq Rehman
  • Matthew P.“Thanks for updating this. I think this makes a really useful tool when talking to the next generation about careers in security being able to demonstrate the breadth of what we can get involved in is quite eye opening”
  • David Elfering“Thank you! I review security programs as part of my work; your outline is fantastic. As a CISO, I regularly reviewed it, and as a consultant/advisor, I can see even more clearly how others would benefit.”
  • Alan Ng“Thanks for all the heart and sweat poured into this awesome mindmap and recommendations! It is a great resource!”
  • Stefan Jäschke – “Terrific work Rafeeq Rehman – thanks for giving back to the community consistently over the years in so many ways.
  • Arvind Javali – “I listened to your podcast interview on CISO mindmap, very insightful, thanks for sharing”
  • Ashoka Reddy – “Thank you, Rafeeq Rehman, for the knowledge, insight, time, and effort you have put into creating and sharing the #CISOMindMap. For #InformationSecurityManagers this is a gold standard that is priceless. It’s a little overwhelming to the point of being funny, but you’ve mapped my professional brain pretty nicely. For #BusinessLeaders this gives oversight to how we protect organizations, revealing our value and responsibilities.
  • Jas Puar“Great timing. I’ve been looking at the 2022 version recently for some inspiration. Glad the 2023 version is available. The role is becoming bigger (cutting deeper across the entire business) and therefore more critical every year. More needs to be done to educate and raise awareness to change the direction of travel, otherwise an already thankless role, will very soon become an impossible one. Keep up the great work!”
  • Steve Lodin – “Thanks again for your continued work here. I look at and keep a copy of every version you release!
  • Fernando Montenegro – “Really nice work, forwarded it to others. I really enjoy the “expiration date” aspect.”
  • Georgeo X. Pulikkathara – “Rafeeq Rehman, good work on the CISO MindMap. My assessment is that this is a good framework for CISOs to approach all the areas we need to consider.
  • Rob Mukherjee – “This is brilliant, thanks Rafeeq. And couldn’t agree more with your comment in the first focus area. “Understand that merely having a backup is not enough. Ability to rebuild impacted systems and restore backups in a timely manner is crucial.” Spot on!! Think “restore”, not just backup!

Acknowledgments

In addition to numerous infosec leaders who provide their input, we have a LinkedIn Group to gather suggestions and comments from the community. While many provided feedback, the following is a list of people and organizations who provided “specific suggestions” for improvements (in no particular order). If I missed anyone, please send me a message to make corrections.

  1. Marc Vael, Esko Chief Digital Trust Officer
  2. Tammy Moskites, CEO & Founder, CISO
  3. Michael Restivo, Vice President | Cybersecurity Executive
  4. Wade Baker Ph.D. Cybersecurity Researcher, Entrepreneur, Professor
  5. Erik Wille, SVP & Chief Information Security Officer Cabinetworks Group 
  6. Kristin Lowery – Vice President, Chief Security Officer at American Electric Power
  7. Omar Khawaja – CISO Databricks
  8. Ross McNaughton, Chief Information Security Officer at Gulf Bank
  9. Rahul Tyagi, Co-founder Safe Security
  10. Andi Baritchi, Vice President, Global Security Assurance, Compliance, Kyndryl
  11. Andres Ricardo Almanza Junco, Cybersecurity Strategist & vBISO | Founder of CISOS.CLUB
  12. Rodrigo Carvalho, Cybersecurity Service Manager, Brazil
  13. Peter Holcomb, ​​CISO | Security Strategist & Advisor
  14. Muath AlHomoud, Director of Cybersecurity, D360 Bank 
  15. Tony DeAngelo, VP, CISO at Encova Insurance
  16. James Azar, Global CISO | Founder X2
  17. Ahmed Kamel, Security ,Risk & Compliance at Nestlé
  18. Ty LeSane, Director of IT Security
  19. Walter Heffel, CISO at ENERSA
  20. Dominik Bredel, Director, Cyber Security bei PwC Deutschland
  21. Darren Kearl, Security Professional @ Siemens Healthineers
  22. Dr. Toh Shang Yee, Head of Information Security at MCIS Insurance Berhad

Following is a list of those (in no particular order) who have contributed to CISO MindMap development in the past.

  1. Jack Jones 
  2. Gary Hayslip
  3. Michael Restivo
  4. Wes Sobbott
  5. Muath AlHomoud  
  6. Ross Young
  7. Ross McNaughton 
  8. Gerard Onorato 
  9. Tony DeAngelo  
  10. James J Azar  
  11. Chris Hughes 
  12. Izhar Mujaddidi
  13. Nadeem Iftikhar 
  14. Ismail Cattaneo  
  15. Andres Ricardo Almanza Junco  
  16. Jack Jones 
  17. Chad Sturgill  
  18. Omar Khawaja  
  19. Rodolphe Simonetti 
  20. Scott Hawk  
  21. Hisham Zahid 
  22. Jerich Beason  
  23. M Kashif Bukhari  
  24. Chris Castaldo
  25. Atif Yusuf
  26. Jon Rogers  
  27. Andi Baritchi 
  28. Ricky Mehra
  29. Ahmed Kamel
  30. Tobias Ander
  31. Indy Dhami
  32. Matthew Thompson
  33. Marc Vael
  34. Christophe Foulon

Your input is highly appreciated!

Older Versions of CISO MindMap

Help Needed

I have heard many recommendations from our community but unfortunately I don’t have enough time to act on these. If you can help in any of the following areas, please reach out.

  1. A dynamic and responsive CISO MinMap page that users can interact with such as clicking on each item and provide further context to become a Wiki page.
  2. Split it out one for a large shop mentality and one for smaller shops.
  3. Add regulatory requirements for different regions.
  4. Translate CISO MindMap into other languages.
  5. An option where one can order a large printed poster.

References

[1] OWASP GenAI Security Project –  OWASP Top for GenAI and LLM

[2] Blog Post – GenAI Risk Categories

[3] Blog Post – How to use GenAI in Cybersecurity Operations

Copyright © note

This MindMap is copyrighted material. However it is absolutely free to all (like water and air) with no strings attached, as long as it is not altered and not used to make money. When using this MindMap, please cite the source properly so that recipients can receive future updates.

Subscribe to Blog

To keep yourself updates and receive email notifications of new posts, use the following link to subscribe to my blog.

Recent Posts

Posted in cisomindmap | Tagged , , , | Comments Off on CISO MindMap 2025: What do InfoSec Professionals Really Do?

How to Use GenAI in Cybersecurity Operations

When GenAI can be used in Cybersecurity programs today and using the tools that are already available? This brief article is to explore some use cases that create immediate value for any Cybersecurity program, small or large.

While there are many, let us briefly list each shown in the diagram above.

  • Security Awareness – Almost all security programs have security awareness as a component. GenAI tools are extremely useful to create content, whether it is an email, brochure or a poster, you can create professional content quite easily decreasing weeks of work into just few hours.
  • Vulnerability Summarization – Imagine you have to provide quick information to your executives about a vulnerability that is being reported in the media. GenAI tools can collect information, summarize, and put it in a nice one-page document quite easily. It is a big time saver!
  • SIEM content development – Just like GenAI is very good at writing Python programs, it is also good at creating detection rules for SIEM, such as Splunk. It is your personal assistant to accelerate new content deployment.
  • Scripting and Automation – Need a quick Python or Shell script? I use these tools to reduce my work from hours to minutes. You can do the same.
  • Code Security Reviews – Code review is made simple with many open source and commercial tools. Many of these tools not only detect logic flaws but also provide recommendations to fix it.
  • Threat Hunting – Feed log data to a model and detect anomalies to give you a starting point for building your hypothesis for threat hunting.
  • Describe a Security Flaw – Instead of writing pages on how common security flaws, such as SQL injection works, get help from GenAI tools.

All people need is a good sense of prompt engineering to use the available tools. Actually learning GenAI will give you many ideas about how you can create more sophisticated applications and agents to carry out more complex tasks in an automated fashion.

Subscribe to Blog

Latest Posts

Posted in AI | Tagged , , , | Comments Off on How to Use GenAI in Cybersecurity Operations

GenAI Risk Categories

Let us make it simple: All GenAI risks can be grouped into three high level risk categories (as explained win this document). These three buckets will help better understand GenAI risks and to apply strategy recommendations below.

Traditional Tech Risks

These risks are associated with network infrastructure on which GenAI models and applications run. These risks always existed and are not specific to AI. These include, but not limited to:

  • Network architecture
  • Operating systems vulnerabilities
  • Applications design
  • Identity and access management
  • Denial of service
  • Web applications (OWASP top 10 for web apps)
  • Ransomware

Recommendation: Continue using traditional controls to manage these risks.

GenAI Amplified Risks

These risks always existed but use of GenAI has either amplified these or created a new dimension to these risk. These include, but not limited to:

  • Data leakage and information disclosure
  • Amplified privacy concerns
  • Global compliance to regulations
  • Incident detection & response
  • Supply chain attacks
  • Resource exhaustion and denial of service

Recommendation: To deal with these risks, security teams need to update and enhance their traditional controls

Newly Introduced Risks

These risks are completely new and very specific to use of GenAI. New methods are needed to identify and manage these risks. These include:

  • Prompt injection
  • Jailbreaking
  • Hallucinations
  • Legal liability of pre-trained models
  • Training data and model poisoning
  • Excessive agency over taking actions

Recommendation: New controls, governance, and tools are needed to manage these newly introduced risks.

Recent Posts

Subscribe to Blog

Posted in AI, Risk Management | Tagged | Comments Off on GenAI Risk Categories

Ten Best Practices for Cybersecurity Risk Management

Cybersecurity risk management is a critical process for organizations aiming to safeguard their assets, systems, and data from potential threats. Effective risk management involves the following best practices: 

  1. Asset Identification : Understand your most critical assets, including data, systems, and networks.
  2. Risk Assessment : Evaluate risks related to each asset by determining potential threats and their likelihood of occurring, as well as the impact on the organization if a threat is realized.
  3. Mitigation Strategies : Develop and implement strategies to mitigate or eliminate identified risks. This may involve technical, administrative, and physical controls.
  4. Continuous Monitoring : Regularly monitor and review the organization’s cybersecurity posture to ensure that new threats are quickly identified and addressed.
  5. Incident Response Plans : Establish incident response plans and procedures to minimize the impact of a successful attack or breach.
  6. Regular Testing and Updates : Regularly test and update your cybersecurity measures to maintain an effective defense against evolving threats.
  7. Security Culture : Foster a security-conscious culture within the organization by providing regular training, encouraging secure habits, and promoting open communication about potential vulnerabilities.
  8. Compliance and Collaboration : Adhere to relevant laws, regulations, and industry standards related to data protection and privacy. Collaborate with other organizations and government agencies to share information about threats and best practices for addressing them.
  9. Patch Management : Maintain a strong focus on patch management to ensure that systems and applications are updated with the latest security patches and fixes.
  10. Threat Intelligence and Vulnerability Management : Utilize threat intelligence frameworks and vulnerability management programs to identify, prioritize, and manage threats based on their likelihood and impact.

By adhering to these best practices, organizations can effectively manage cybersecurity risks, protect their assets, systems, and data, and reduce the likelihood of a security breach. 

Posted in cisomindmap, InfoSec | Tagged , | Comments Off on Ten Best Practices for Cybersecurity Risk Management

Run LLM Models on Macbook – Part II

Protect your intellectual property by running GenAI models locally

This is a continuation of my previous blog post on running open source models locally. In this blog we add Open WebUI as a web interface to provide the end user similar experience as ChatGPT.

Following is a short summary of why run local models locally.

Why Run Open-Source GenAI Models Locally?

Running open-source GenAI models locally provides several benefits:

  1. Data Protection: Keep sensitive information within your organization’s control. By processing data locally, you can keep sensitive information from leaving your organization’s premises, reducing the risk of data breaches and intellectual property theft.
  2. Increased Control: With full control over the deployment environment, you can tailor the setup to meet your specific requirements, ensuring that your models are used as intended.
  3. Improved Performance: Running GenAI models locally allows for faster processing times and reduced latency, making them more suitable for real-time applications.
  4. Enhance Security: Reduce the risk of data breaches and intellectual property theft by processing data locally.

Open Source License Requirements

Please check the licensing requirements for the open source model you are going to use as it may be quite different depending upon the model you use. The license may also be different depending on the type of use (personal, educational, commercial, etc.)

Step-by-Step Process

Follow these three simple steps to install and run multiple LLMs locally.

Step 1: Set up Ollama

Ollama is an open source project that enables you to run large language models (LLMs) locally without going through too much hassle. It is available at GItHub (see references below)

Open terminal window on Macbook and use the following commands:

mkdir llm
cd llm
curl -L https://ollama.com/download/ollama-darwin-arm64 -o ./ollama
chmod u+x ollama 
./ollama

Step 2: Download LLM model you want to run

Go to a new terminal windows and pull llama3

./ollama pull llama3

Step 3: Run the Model

Once the model is downloaded, you can run it and use prompts on the command line. Following is a typical session with one prompt and its response:

./ollama run llama3

Step 4: Install Open WebUI

Use the following steps to install Open WebUI.

python3 -m venv venv
 . venv/bin/activate  
pip install open-webui

The first line creates a new Python environment. The second line activates this environment and the third line installs Open WebUI.

Step 5: Run Open WebUI

Use the following command to run Open WebUI.

open-webui serve

You will see a bunch of messages appear on the terminal window as shows in the screenshot below. 

Now point your web browser to http://0.0.0.0:8080 or http://localhost:8080 where you have to create an account the first time you use Open WebUI. You are ready to ask questions using this web interface. On the top-left corner of the following screenshot, select one of the installed models from step 2 and then enter prompt for that model.

If you have multiple models installed locally, you can use the same prompt to check how they respond differently.

References

Subscribe to Blog

Subscribe to blog to get email notification of new posts

Recent blog posts

Posted in AI | Tagged , , , | Comments Off on Run LLM Models on Macbook – Part II

Post-Quantum Cryptography Resources

Open Source Software Libraries and Organizations working to future proof security of data 

This blog post provides essential resources for security professionals and software developers looking to secure data in the post-quantum era. It highlights key organizations such as NIST, the Linux Foundation, and the Post-Quantum Cryptography Alliance, which are leading efforts in quantum-resistant cryptography. Additionally, it introduces open-source libraries that facilitate the implementation of quantum-safe encryption algorithms, offering practical tools for integrating cutting-edge security measures into your systems.

NIST Post-Quantum Cryptography Research

The NIST Post-Quantum Cryptography project focuses on developing and standardizing cryptographic algorithms that are secure against the threats posed by quantum computing. This initiative aims to identify and endorse new cryptographic algorithms that can protect data and communications in a future where quantum computers might break current encryption methods. The project involves a multi-phase process, including a public evaluation and selection process to ensure that the chosen algorithms are both secure and practical for widespread use. NIST’s goal is to establish new standards for cryptography that will ensure data security well into the future.

Main webpage: https://csrc.nist.gov/projects/post-quantum-cryptography

NIST Released Three Post-Quantum Encryption Standards

After 8 years of work, the National Institute of Standards and Technology (NIST) has released its first three finalized post-quantum encryption standards, marking a significant step in securing data against future quantum computing threats. These standards include algorithms designed to protect digital information from being compromised by quantum computers, which have the potential to break traditional encryption methods. The new standards are intended to ensure long-term data security and support the transition to quantum-resistant cryptographic systems. 

Main webpage: The announcement page

Following are links to the finalized standards:

OpenQuantum Safe

The Open Quantum Safe (OQS) project is dedicated to developing and promoting quantum-resistant cryptographic algorithms and tools to safeguard data against the potential threats posed by quantum computing. It provides an open-source platform for the research and implementation of cryptographic algorithms that are resistant to quantum attacks. The OQS project aims to advance the security of information systems by offering resources, software libraries, and a community-driven approach to integrating quantum-safe algorithms into existing security frameworks.

Post Quantum Cryptography Alliance

The Post-Quantum Cryptography Consortium (PQCA) is focused on advancing the adoption and development of cryptographic systems that are secure against the potential threats posed by quantum computing. The consortium collaborates with industry leaders, researchers, and institutions to facilitate the transition to quantum-resistant cryptographic algorithms. It provides resources, guidelines, and support to help organizations implement and evaluate post-quantum cryptographic solutions, ensuring robust security in a future where quantum computers could compromise current encryption methods.

Open Quantum Safe provider for OpenSSL

The OQS Provider on GitHub is an open-source project that integrates quantum-safe cryptographic algorithms into the Open Quantum Safe (OQS) framework, such as OpenSSL. It provides a modular interface for incorporating post-quantum cryptographic algorithms into existing cryptographic libraries and systems. The OQS Provider enables developers to test and deploy quantum-resistant algorithms by offering an implementation of these algorithms as a provider for cryptographic operations. The project aims to facilitate the evaluation and adoption of quantum-safe cryptography within various software environments.

Summary

The post highlights essential resources for security professionals and software developers to secure data in the post-quantum era. It mentions key organizations such as NIST, the Linux Foundation, and the Post-Quantum Cryptography Alliance (PQCA), which are leading efforts in quantum-resistant cryptography. The blog also introduces open-source libraries that facilitate the implementation of quantum-safe encryption algorithms, including Open Quantum Safe (OQS) and its provider for OpenSSL. Specifically, it notes NIST’s release of three post-quantum encryption standards, FIPS 203-205, and provides links to these standards. Additionally, it summarizes OQS as an open-source project dedicated to developing and promoting quantum-resistant cryptographic algorithms and tools, and PQCA as a consortium focused on advancing the adoption and development of cryptographic systems secure against quantum threats.

Subscribe to Blog

Posted in AI, InfoSec, Quantum | Comments Off on Post-Quantum Cryptography Resources

Run LLM Models on a Macbook

As the use of GenAI models becomes increasingly prevalent, it’s crucial for organizations to ensure the security and ownership of their intellectual property. One way to achieve this is by running open-source GenAI models locally on your own infrastructure.

Why Run Open-Source GenAI Models Locally?

Running open-source GenAI models locally provides several benefits:

  1. Data Protection: Keep sensitive information within your organization’s control. By processing data locally, you can keep sensitive information from leaving your organization’s premises, reducing the risk of data breaches and intellectual property theft.
  2. Increased Control: With full control over the deployment environment, you can tailor the setup to meet your specific requirements, ensuring that your models are used as intended.
  3. Improved Performance: Running GenAI models locally allows for faster processing times and reduced latency, making them more suitable for real-time applications.
  4. Enhance Security: Reduce the risk of data breaches and intellectual property theft by processing data locally.

Open Source License Requirements

Please check the licensing requirements for the open source model you are going to use as it may be quite different for each model. The license may also be different depending on the type of use (personal, educational, commercial, etc.)

Step-by-Step Process

Follow these three simple steps to install and run multiple LLMs locally.

Step 1: Set up Ollama

Ollama is an open source project that enables you to run large language models (LLMs) locally without going through too much hassle. It is available at GItHub (see references below)

Open terminal window on Macbook and use the following commands:

mkdir llm
cd llm
curl -L https://ollama.com/download/ollama-darwin-arm64 -o ./ollama
chmod u+x ollama 
./ollama

Step 2: Download LLM model you want to run

Go to a new terminal windows and pull llama3

./ollama pull llama3

Step 3: Run the Model

Once the model is downloaded, you can run it and use prompts on the command line. Following is a typical session with one prompt and its response:

./ollama run llama3
>>> who wrote harry potter?
>>>

If you have completed the above three steps, you are on your way to start using LLMs locally on your Macbook. No need to have Internet connectivity or send data to a third party chatbot. Ollama supports many LLMs and the list is available on its Github page. You can “pull” any of these models and test them locally.

Running open-source GenAI models locally provides a secure and controlled environment for developing and deploying AI-powered applications. By maintaining ownership and control over your data, you can protect your intellectual property while still leveraging the benefits of GenAI technology.

In a next blog post, I will discuss using a web interface with Ollama.

References

https://github.com/ollama 
https://github.com/ollama/ollama
https://ollama.com/library

Subscribe to My Blog

Latest Blog Posts

Posted in AI | Comments Off on Run LLM Models on a Macbook

Navigating the Landscape of Risk Management Frameworks

In the realm of information security, the quest to effectively manage risk is paramount. However, amidst the myriad of frameworks available, distinguishing between those explicitly designed for risk management and those that serve as broader guidelines can be a daunting task. A recent survey conducted over a 24-hour period on April 20-21, 2024, offers illuminating insights into the landscape of risk management frameworks, shedding light on prevalent practices and perceptions within the industry. Following is an image showing the results of this survey:

Despite the survey question’s slight ambiguity, the responses gleaned from 259 votes and 5591 impressions on LinkedIn provide valuable glimpses into the strategies employed by information security teams. It becomes evident that a multitude of frameworks are being utilized, even ones not originally intended as strict “risk management frameworks”.

Towering above the rest in popularity is the NIST Cybersecurity Framework (NIST CSF), a stalwart in the field renowned for its comprehensive approach to cybersecurity. Originating from an initiative aimed at safeguarding critical infrastructure, NIST CSF has transcended its initial scope to emerge as a go-to resource for managing cybersecurity risk across diverse organizational landscapes.

When probed further about why individuals perceive NIST CSF as a risk management framework, one respondent succinctly encapsulated the sentiment: “The goal of CSF and pretty much any control framework is to manage and reduce risk.” Indeed, the preface of NIST CSF explicitly states its purpose as assisting organizations in managing and mitigating cybersecurity risks, underscoring its relevance in the risk management arena.

However, it’s important to acknowledge dissenting voices within the survey responses. While many recognize NIST CSF as a formidable tool for risk management, others contend that it falls short of being a dedicated risk management framework. Instead, frameworks such as ISO/IEC 27005:2022 and NIST SP800-30 are hailed as the true champions of risk management, offering more specialized approaches tailored to the intricacies of risk assessment and mitigation.

Moreover, respondents highlight the need to delineate between distinct categories of frameworks, including:

  • Security Program Management such as NIST CSF
  • Risk Management such as NIST RMF
  • Control Frameworks (such as NIST 800-53)

This clarification is essential in preventing confusion and ensuring that organizations select the most appropriate framework for their specific needs.

Further complexity arises with the mention of FAIR (Factor Analysis of Information Risk), heralded as a risk assessment methodology rather than a traditional risk management framework. While some dismiss FAIR due to its perceived lack of widespread adoption, others emphasize its value in providing a structured approach to assessing and quantifying information risk.

In conclusion, the survey findings underscore the nuanced landscape of risk management frameworks, characterized by a diversity of approaches and perspectives. While NIST CSF reigns supreme in popularity and utility, it’s essential for organizations to critically evaluate their needs and objectives before selecting a framework (especially when it comes to SEC breach notification/reporting on Form 8K). By fostering a deeper understanding of the distinctions between various frameworks and methodologies, information security teams can navigate the complex terrain of risk management with confidence and clarity.

Subscribe to blog

Recent Posts

Posted in Risk Management | Tagged , , , | Comments Off on Navigating the Landscape of Risk Management Frameworks

CISO MindMap 2024: What do InfoSec Professionals Really Do?

Many individuals outside the realm of cybersecurity often underestimate the intricacies involved in a security professional’s role. Since its inception in 2012, the CISO MindMap has served as a valuable educational resource, offering insights into CISO responsibilities and aiding security professionals in crafting and enhancing their security programs. Continuously adapting to reflect the evolving landscape of cybersecurity, the CISO MindMap has been updated to accommodate the latest developments in the field. Here is the most recent iteration of the CISO MindMap for 2024, featuring numerous enhancements and fresh recommendations for the year 2024-25.

Don’t forget to review recommendations for 2024-25 described below and to subscribe to my blog.

Summary of Changes to 2024 CISO MindMap

With time, the responsibilities of security professionals are only increasing. Why? Technology is changing fast, bringing new ways of doing business, continuous adoption of Cloud, and extremely fast evolving GenAI technology with many applications. Not only the Infosec professionals are “expected” to deeply understand these technologies, they are also tasked with providing policies/guidance on how to secure them. For this reason, every year you find new things on the CISO MindMap. At the same time, some items are changed or removed from the CISO MindMap depending upon their relevance. In the latest CISO MindMap, modified and new items are marked in red color for your convenience.

Other noticeable changes include:

  • Artificial Intelligence and GenAI – MindMap now has a dedicated section on Artificial Intelligence and GenAI, reflecting insights from security leaders. GenAI’s rapid development is exciting, but measured caution is advised until the field matures further. 
  • Removing Redundancy and Overlaps – Removed some redundancies and overlaps in different sections. I also moved some elements of the MindMap to more relevant categories.
  • Security Team Branding – An integral aspect of the CISO’s role involves effectively advocating for information security across diverse stakeholders. Drawing from valuable insights provided by experienced CISOs, we have expanded the section on ‘Security Branding’ in response to their feedback. This includes essential skills such as negotiation, executive engagement, strategic prioritization, and tactful decision-making. These are identified as key areas where new CISOs often encounter challenges, and thus, warrant special attention.
  • Expiration Date – A common issue is that many professionals still have older CISO MindMap copies. Like last year, I added an “expiration date” to let people know when they should stop using a particular version. The expiration date for the 2024 CISO MindMap is the end of June 2025. The next version will be published before the current version expires.

CISO MindMap Updates Methodology

Every update to the CISO MindMap undergoes thorough consideration, research, and attention to detail. In addition to my ongoing engagements with industry leaders, various methods are employed to ensure we capture the pulse of the cybersecurity landscape:

  • Conducting interviews with experts
  • Distributing surveys to gather insights
  • Leveraging LinkedIn for targeted questions and discussions
  • Analyzing feedback and comments from previous versions of the CISO MindMap
  • Staying abreast of industry news and conducting in-depth analysis

Furthermore, I’d like to express my gratitude to the contributors whose valuable insights have enriched this endeavor. The ‘Acknowledgments‘ section of this blog post includes their names and LinkedIn profiles as a token of appreciation.

Are you Accountable for Everything in the CISO MindMap?

Security is inherently a collaborative effort, and the role of the CISO entails providing consultative guidance in various areas outlined within the CISO MindMap. It’s crucial to discern between areas where direct ownership and accountability are necessary and those where consultation is the primary function. Within any organization, numerous stakeholders are involved, and a common pitfall is the lack of clearly defined boundaries for each role.

My recommendation is to establish a RACI (Responsible, Accountable, Consulted, Informed) matrix, which serves as a standardized methodology for delineating roles and responsibilities among stakeholders [6]. By mapping out tasks and corresponding roles within this framework, clarity is achieved, ensuring that each stakeholder understands their level of involvement and contribution to the overall security landscape.

Focus Areas and Recommendations for 2024-2025

Each year, I offer my recommendations as a practitioner, drawing insights from discussions with information security leaders. My approach strives for objectivity, steering clear of hype and focusing solely on data-driven research. Though unintended biases may exist, the goal remains to propose actionable steps viable within a short to mid-range timeframe. These recommendations do not constitute future predictions; rather, they address the immediate needs for enhancing security programs.

Selecting a concise set of recommendations is always a challenge for me. While the list provided below may be longer than I’d prefer, I aim to offer a comprehensive array of considerations for your reflection. Followers of the MindMap will notice both recurring suggestions from previous years and novel additions, providing a blend of continuity and fresh insights.

I’m genuinely interested in hearing your perspective on these recommendations and understanding whether they resonate with your experiences and insights. Your feedback, whether in agreement or disagreement, provides valuable insights that can help refine and improve our approach to addressing security challenges. So, please feel free to share your thoughts and insights on these recommendations—whether you support them or have reservations—so we can engage in a constructive dialogue to further enhance our security strategies.

Recommendation 1: Adopt a Cautious Approach Towards GenAI

GenAI has become a focal point of discussion, rapidly evolving within the technological landscape. The Wall Street Journal reported Amazon’s unprecedented investment in an AI startup, reflecting the significant momentum in this domain. In 2023, investments in Generative AI companies exceeded $29 billion [1]. Furthermore, numerous open-source solutions are emerging within the GenAI market [2]. Amidst the buzz, it’s important to maintain a discerning approach.

Security professionals approach GenAI from three primary perspectives:

  1. Utilization within security practices: Emphasizing enhanced productivity and the exploration of automation opportunities for routine tasks.
  2. Safeguarding GenAI in corporate environments: Establishing GenAI centers of excellence to ensure governance, policy adherence, and protection against potential attacks. Aligning business and cybersecurity strategies to accommodate GenAI.
  3. Monitoring malicious use of GenAI: While not yet a cause for alarm, it’s essential to remain vigilant regarding the evolving applications of GenAI for malicious purposes.

Although every security vendor touts AI integration within their products and services, it’s prudent to avoid succumbing to hype and refrain from significant investments in GenAI technology for security purposes at this stage. It’s anticipated that over the next 12-18 months, clear market leaders will emerge from the current phase of hype and use cases for Cybersecurity will be better defined. At this stage, it is better to educate oneself and explore use cases such as automation and productivity enhancement.

Recommendation 2: Consolidate and Rationalize Security Tools

No matter the size of your InfoSec budget, it is prudent to take a more critical look at security programs and find ways of ways for program management. An average organization is using a large number of technologies, by some estimates as high as 47, according to a Ponemon survey. Yet many security leaders don’t know if Cybersecurity tools are working. Accumulating more security tools doesn’t necessarily lower risk; rather, it amplifies the necessity for maintaining expertise within security teams. There is a need for consolidation and rationalization of security tools by deeply exploring Return on Investment (ROI) of these tools. When rationalizing the need for tools, consider factors such as functionality overlap, available expertise within the team to effectively utilize the tool, and the innovation on the side of vendors. You may be surprised to find shelfware. In some cases open source tools may work just fine as well.

Recommendation 3: Cyber Resilience – Go Beyond Incident Response

Building upon last year’s recommendation, it’s evident that in numerous organizations, security incident response remains separate from business continuity and disaster recovery functions. It’s imperative to assess ransomware defenses, detection, and response capabilities comprehensively. Conducting a business impact analysis is essential to identify critical processes, applications, and data.

Moreover, testing the ability to restore systems and data within an acceptable timeframe is crucial. Merely possessing backups is insufficient; the capability to rebuild impacted systems and restore backups promptly is vital for restoring normal business operations following security incidents.

Recommendation 4: Build a Brand for Security Team

The role of a CISO is inherently public-facing, demanding continuous engagement with technology and business leaders. While the message being conveyed holds significance, the credibility of the messenger is equally vital. To better serve the interests of the business, it’s imperative to equip security team members with skills beyond technical expertise, especially those who are aspiring to be leadership roles in future.

This includes training in business acumen, value creation, influencing without authority, and enhancing human experience. Emphasizing these aspects is not new; however, it’s crucial to maintain focus on the fact that information security teams operate within a broader context and must facilitate business objectives while collaborating effectively with others.

Skills such as negotiation, compromise, and strategic decision-making play pivotal roles in establishing both personal and team credibility, ultimately contributing to the enhancement of the overall brand.

Recommendation 5: Maximize Business Value of Security Controls

The proliferation of security and compliance controls often creates friction between security and other technology teams. However, not all controls carry equal weight, and some may offer minimal value in terms of risk reduction and business enhancement. It is advisable to conduct a thorough assessment of each control, evaluating its effectiveness and business impact.

CISOs should devise a strategy to phase out low-value controls within the next 12-18 months, thereby eliminating unnecessary maintenance costs. This proactive approach ensures resources are allocated efficiently, focusing efforts on controls that truly contribute to mitigating risks and driving business value

How to Use CISO MindMap?

Have you ever been asked to explain what you do as a security professional? The CISO MindMap offers a comprehensive solution for addressing this question and clarifying the intricacies of the role. Many professionals have attested to its efficacy in elucidating the complexities of the CISO position, particularly when communicating with a business audience.

Here are some ways in which the CISO MindMap proves to be immensely valuable:

  • Facilitating conversations with fellow technology professionals.
  • Instrumental in the design and refinement of security programs.
  • Adopted by certain security vendors for raising awareness.
  • Employed in CISO group discussions and community meetings.
  • Aiding aspiring security professionals in understanding the industry landscape and charting their career paths.
  • Serving as an educational and awareness-raising tool.

Obviously there is a lot on this MindMap. The stress on people who have these responsibilities is real. If nothing else, this MindMap should help leaders recognize that stress and do something about it. I covered this topic (stress) in my latest book Cybersecurity Arm Wrestling: Winning the perpetual fight against crime by building a modern Security Operations Center (SOC) as well.

What are They Saying?

Following are some comments on CISO MindMap LinkedIn post from 2023.

  • Michael Restivo“One of the most valuable documents around. Love this Rafeeq.”
  • Chris Novak“Always a great asset! Thanks for all that you do for the cybersecurity community!”
  • Muath AlHomoud“Great insight Rafeeq Rehman always inspiring”
  • Christophe Foulon “Insightful as always, Rafeeq Rehman
  • Matthew P.“Thanks for updating this. I think this makes a really useful tool when talking to the next generation about careers in security being able to demonstrate the breadth of what we can get involved in is quite eye opening”
  • David Elfering“Thank you! I review security programs as part of my work; your outline is fantastic. As a CISO, I regularly reviewed it, and as a consultant/advisor, I can see even more clearly how others would benefit.”
  • Alan Ng“Thanks for all the heart and sweat poured into this awesome mindmap and recommendations! It is a great resource!”
  • Stefan Jäschke – “Terrific work Rafeeq Rehman – thanks for giving back to the community consistently over the years in so many ways.
  • Arvind Javali – “I listened to your podcast interview on CISO mindmap, very insightful, thanks for sharing”
  • Ashoka Reddy – “Thank you, Rafeeq Rehman, for the knowledge, insight, time, and effort you have put into creating and sharing the #CISOMindMap. For #InformationSecurityManagers this is a gold standard that is priceless. It’s a little overwhelming to the point of being funny, but you’ve mapped my professional brain pretty nicely. For #BusinessLeaders this gives oversight to how we protect organizations, revealing our value and responsibilities.
  • Jas Puar“Great timing. I’ve been looking at the 2022 version recently for some inspiration. Glad the 2023 version is available. The role is becoming bigger (cutting deeper across the entire business) and therefore more critical every year. More needs to be done to educate and raise awareness to change the direction of travel, otherwise an already thankless role, will very soon become an impossible one. Keep up the great work!”
  • Steve Lodin – “Thanks again for your continued work here. I look at and keep a copy of every version you release!
  • Fernando Montenegro – “Really nice work, forwarded it to others. I really enjoy the “expiration date” aspect.”
  • Georgeo X. Pulikkathara – “Rafeeq Rehman, good work on the CISO MindMap. My assessment is that this is a good framework for CISOs to approach all the areas we need to consider.
  • Rob Mukherjee – “This is brilliant, thanks Rafeeq. And couldn’t agree more with your comment in the first focus area. “Understand that merely having a backup is not enough. Ability to rebuild impacted systems and restore backups in a timely manner is crucial.” Spot on!! Think “restore”, not just backup!

Acknowledgments

In addition to numerous infosec leaders who provide their input, we have a LinkedIn Group to gather suggestions and comments from the community. While many provided feedback, the following is a list of people and organizations who provided “specific suggestions” for improvements (in no particular order). If I missed anyone, please send me a message to make corrections.

  1. Jack Jones 
  2. Gary Hayslip
  3. Michael Restivo
  4. Wes Sobbott
  5. Muath AlHomoud  
  6. Ross Young
  7. Ross McNaughton 
  8. Gerard Onorato 
  9. Tony DeAngelo  
  10. James J Azar  
  11. Chris Hughes 
  12. Izhar Mujaddidi
  13. Nadeem Iftikhar 
  14. Ismail Cattaneo  
  15. Andres Ricardo Almanza Junco  
  16. Jack Jones 
  17. Chad Sturgill  
  18. Omar Khawaja  
  19. Rodolphe Simonetti 
  20. Scott Hawk  
  21. Hisham Zahid 
  22. Jerich Beason  
  23. M Kashif Bukhari  
  24. Chris Castaldo
  25. Atif Yusuf
  26. Jon Rogers  
  27. Andi Baritchi 
  28. Ricky Mehra
  29. Ahmed Kamel
  30. Tobias Ander
  31. Indy Dhami
  32. Matthew Thompson
  33. Marc Vael
  34. Christophe Foulon

Your input is highly appreciated!

References

[1] Wall Street Journal, Amazon Invests $2.7 Billions More in AI Startup, [link

[2] Wall Street Journal, Open-Source Companies Are Sharing Their AI Free. Can They Crack OpenAI’s Dominance?, [link]

[3] NIST AI Risk Management Framework [link]

[4] Deloitte, Proactive Risk Management in Generative AI [link]

[5] Deloitte, The implications of Generative AI for businesses [link]

[6] CIO, The RACI matrix [link]

Copyright © 2012-2024 – Permission to use

This MindMap is copyrighted material. However it is absolutely free for all personal, business and professional purposes (like water and air). There are no strings attached, as long as it is not altered and not used to make money. When using this MindMap, please cite the source properly so that recipients can receive future updates.

Subscribe to Blog

To keep updated with new versions of CISO MindMap and other posts, subscribe to my blog here:

Recent Posts

Posted in cisomindmap | Tagged , , , , , | Comments Off on CISO MindMap 2024: What do InfoSec Professionals Really Do?

Building Generative AI (GenAI) Applications

Four key considerations for business executives

Harnessing the potential of Generative AI (GenAI) to create user applications that drive business value may appear daunting, yet it doesn’t need to be. With the GenAI field advancing swiftly and offering a plethora of options, understanding the following four fundamental areas can provide clarity for business leaders:

  1. User Interface (UI) Design for Prompt Interaction – GenAI applications typically rely on user prompts to generate responses. A well-designed UI, effective prompt engineering, rigorous testing, and user training are crucial aspects. Tailoring the UI to enable users to establish “context” for GenAI responses is paramount.
  1. Implementing Prompt and Response Filtering – Establishing controls to govern how GenAI models respond to prompts and filtering out undesirable content is essential. This ensures that the generated responses align with organizational standards and objectives. The focus should be on providing guardrails in case the GenAI models deviate from safety standards.
  1. Selecting a Baseline GenAI Model – Choosing the right baseline GenAI model is pivotal. These pre-trained deep learning models serve as the foundation and offer a starting point for application development. Careful consideration of factors such as application type, cost, and computational requirements is imperative when selecting from a range of commercial and open-source options.
  1. Fine-Tuning for Optimal Performance – Enhancing the performance of baseline models through fine-tuning is indispensable. Fine-tuning involves leveraging additional data to adapt the baseline model to specific use cases, such as generating journalistic essays, customer support, legal documents, etc. Organizations should deliberate on how they intend to fine-tune GenAI models to align with their unique requirements.

While building GenAI applications entails inherent complexity, focusing on these considerations provides a valuable blueprint for developing high-quality user applications driven by Generative AI technology.

Subscribe to Blog

Recent Posts

Posted in AI, Leadership | Tagged , , | Comments Off on Building Generative AI (GenAI) Applications