First of all, I am not abusing AI 🤐
WTF means “What the Funn is AI”. [My slag] 😜
From today, we will start our AI Security journey.. I will try to cover a maximum of topics which come under AI Security.
As today is our first day, let’s start with the basics in a simple and short way.
1.What is Artificial Intelligence
Even a 5 year kid knows AI such as chatgpt, gemini etc. But very few people truly understand what it is exactly, how it works and why it can be dangerous if misused?
In simple terms, AI is the ability of machines to perform tasks that normally requires human intelligence.
Usually it tries to mimic us [human].
These tasks include:
- Recognizing images
- Understanding language
- Making decisions
- Learning from past data
2.Traditional Software vs AI
Many of us have a myth that AI is just an extended version of Normal Software. But it’s not.
Traditional Software
Traditional Software usually follows fixed rules which is entirely written by humans
AI
Imagine AI as a kid, It learns from its parents (the data) and tries to speak through random sounds and words.
AI learns from the data, and the rules are not explicitly written, just as a child has no restrictions and learns in their own way.
#Note: Because AI learns from data, anyone who controls or manipulates that data can influence the AI’s behavior.
3. AI vs Automation
Another myth is many people think automation and AI are both the same. [Automation = AI]
But, they are not.
Automation
Automation just executes predefined steps, no learning process is involved and no intelligence is used.
Ex : Automatic email reply
AI
AI learns from the patterns, makes predictions and improves with more data.
Ex : Email spam detection
Why Does Security Matters?
Automation can only fail if code is wrong.
But AI can fail even when the code is correct because :
- Data can be poisoned
- Models can be manipulated
- Output can be tricked
4. Narrow AI and General AI
There are 7 types of AI including Narrow AI, General AI, Superintelligent AI, Reactive Machines, Limited Memory AI, Theory of Mind AI, Self-Aware AI. but as our main focus is on AI Security
Narrow AI [Weak AI]
This is all the AI we have today, like ChatGPT, Voice assistants etc.
Characteristics:
- Designed for one specific task
- Cannot think outside its scope
- Very powerful but very limited
Security risk:
- Easy to exploit with crafted inputs
General AI [Strong AI]
This is theoretical and still in the research phase, like a machine can learn medicine, law, art and security by itself.
Characteristics:
- Human level intelligence
- Can reason across domains
- Can learn anything like human
Security risk:
- If General AI ever exits, security becomes a global concern.
5. Why can AI be Misused?
At present we are in the AI era, AI is everything, AI is GOAT, AI is our BFF..
Just like every happiness comes with some limitations. The risk comes from how AI is used or abused.
Few Common Misuses:
- Generating fake images, videos and voices
- Automating phishing and scams
- Spreading misinformation
- Bypassing security systems
Ex : Using AI a single attacker can now perform the work of thousands of attackers easily.
6. Key Takeaways
- AI learns from data, not rules.
- AI ≠ Automation
- All current AI is Narrow AI
- AI is everywhere, often invisible
- AI can be misused at massive scale
- AI security will soon become a global concern.
Join Hacklido Telegram Channel to get notified for latest blogs, security tips, tools, latest news and more✨
https://t.me/hacklido
