Cyberattacks are our present and future

What makes this channel interesting is how natural it is. You may have come here for different reasons, but you read it because I describe my own experience of building AI projects. There are no news posts or AI-written posts here. Just the personal views of a practitioner and entrepreneur.

Yesterday, a vulnerability appeared in PHP, a widely used programming language for websites. Websites around the world written in this language could be hacked. If you are interested, you can read the details here: https://www.bbc.com/russian/articles/cwe8lz24pprjo

What I realised:

The bigger and more complex the internet becomes in its structure and capabilities, the more holes it has. In the Development 2.0 podcast, I said that, contrary to the view that AI will replace developers, the opposite will happen. There will be many more developers, and all of them will use AI.

This creates fertile ground for hacks and breaches aimed at blackmail, data theft, espionage or even just amusement.

We cannot ignore this, so I realised I need to focus on cybersecurity in my projects, U2234.com and CrashBrief.com.

I do not yet know exactly how this will work. Perhaps some kind of AI guard running 24/7/365, perhaps software that takes a service offline when an alarm goes off, perhaps something else. We will also need to work through the project with all the agents to minimise opportunities for attacks and security holes.

It is easier to do this now, at the beginning, while the projects are still easy to change and manage. For large companies it may be much harder. Security is an ongoing compromise between convenience and protection. Every new security layer adds another action for the user, and the human factor plays a very significant role, perhaps the decisive one.

You can give an agent clear instructions with 105 steps and it will follow them without breaking a sweat. A person will not.

I see both great potential for work and a great challenge here, because it will involve a lot of interaction with people who cannot be cyborgs. Agents will need to take on most of the work.