AI & Non-AI Data

AI has created a good percentage of the content that the internet is showing. I was wondering if there was a number and the best guess is around 50%


So, half of what you may see on the internet may be AI-based and not human-based. I do not know if this is a problem or if it is simply how data is going to work. I think the problem is that AI content is not always the most reliable. Some organizations do checks and balances. For others, it is literally the wild, wild West. This is not a good thing because data is treated as currency and when that is diluted there are big problems that arise. 


I recently read about Nightshade being used to protect the work of artists. It changes the pixels and causes AI to think that the wrong information is really something that it is not. Say there is a cat and the work used Nightshade and AI may now think that the cat is a flower and when a user requests a flower the return no longer will be a true flow, but rather, a cat feature flower. Is there a good reason to protect intellectual property? Yes. It is also destroying AI-created content and the more diluted the content the more diluted and wrong responses people are going to get when using AI. Double-edged sword. Good and bad all at the same time. 


Now, would AI still come to this same problem without the use of NIghtshade and other protection programs?


In my opinion, probably. I think about the various answers that I can get from AI when I ask the same question in different ways. 


I can ask about investing and what stocks to use and may get a response that it is not allowed to provide investment advice. I can rephrase those questions and ask what the highest-yielding stocks are for the year and receive a very specific response. I can also ask what stocks various people have invested in and what their returns are and get a full answer. This means that there are safeguards for AI except the problem is that you only know how to secure something after it has been broken. 


Often I have to think about all the scenarios that might happen in a given environment and do my best efforts at protecting those scenarios. The problem is that I cannot think of every single scenario and situation that may arise and often I go from being on the offensive to being on the defensive because I simply do not know how it is going to be attacked. 


I think about this when I see a warning label on a particular product and try to figure out what someone had to do to make that company put that label on there. They are not putting “not for internal use” on curling irons for no reason…


The world becomes more and more interesting every day and I learn new things about the world. Some good and some bad. I am not sure where AI will take us. All I know is that it will continue to stay in our daily lives more than we realize because it can run behind the scenes in many of the applications we use and simply would not know.


The more AI is used. The more of a need to ensure that there are regulations to protect the data it processes. We just started making privacy regulations in the United States and I do not think that the legal system is ready to fast-forward that process and include AI in the current year.


It will be interesting to see what the different solutions are to keep data protected and information from being diluted over the next year.


Comments

Post a Comment

Popular posts from this blog

Understanding Protection (SC)

Understanding Protection (SC) These controls focus on protecting system boundaries, communications, and preventing unauthorized data exposure. Let's break down the key SC controls with examples. SC-1: Policy and Procedures This foundational control requires documented policies and procedures for system and communications protection. Example: SYSTEM AND COMMUNICATIONS PROTECTION POLICY Version: 2.1 Last Updated: 2025-01-02 1. PURPOSE This policy establishes requirements for protecting the organization’s cloud infrastructure and communications. 2. SCOPE Applies to all cloud systems within the authorization boundary. 3. POLICIES 3.1 Encryption Requirements - All data in transit must use TLS 1.2 or higher - All data at rest must use FIPS 140-2 validated encryption - Key rotation required every 365 days 3.2 Network Security - All external connections must traverse a DMZ - Firewall rules follow deny-by-default principle - Monthly review of access control lists 4. PROCEDURES 4.1 Firewall ...
Read more

Security Assessment (SA) Controls

The Security Assessment (SA) family of controls, derived from NIST 800-53 Revision 5, plays a pivotal role in keeping systems secure over time. What Are the Security Assessment (SA) Controls? The SA family of controls focuses on ensuring that security controls are assessed for their effectiveness and are continuously monitored throughout the life of the system. This means checking whether the security measures in place are not only effective but remain effective and up-to-date as time progresses. For FedRAMP Moderate, these controls are crucial because: They ensure security controls are evaluated regularly. They support continuous monitoring for vulnerabilities or weaknesses. They establish corrective actions when necessary. SA-1: Security Assessment and Authorization Policies and Procedures Control Overview SA-1 requires the development of security assessment and authorization (A&A) policies and procedures. These procedures outline the process for conducting security assessm...
Read more

FedRAMP Moderate Rev 5 SI Controls

  In this post, I'll break down each SI control in FedRAMP Moderate Rev 5 and provide tips.   SI-1: System and Information Integrity Policy and Procedures The key to a strong base is well-documented policies and procedures.  For SI-1, organizations must: - Develop and maintain system and information integrity policies that address purpose, scope, roles, and responsibilities - Define procedures for implementing security controls - Review and update these documents at least annually Example:  Create a System and Information Integrity Policy document that includes: - Malware protection requirements with procedures - System monitoring procedures - Software and firmware update processes - Security alert handling processes and procedures - Error handling protocols  SI-2: Flaw Remediation Flaw remediation is key for maintaining system security.  Organizations must: - Identify, report, and correct system flaws promptly - Test software updates before deployment...
Read more