Dangers of AI: Can we ever eliminate them?

I want to preface this article with my love of the technology, my belief in it’s innovation, and desire to see it strive to its potential. I have written articles about the subject prior to and upon the release of beta OpenAI GPT. I continue to strive to bring the latest on the subject and champion it in the greater world beyond that of just the AI and Technology Communities. With people’s content like Vin Vashishta , Andreas Welsch , Ronald van Loon , Nicolas Babin , and Tony Moroney we are seeing its integration into business, marketing and finance. We of course are also seeing it other fields from programming, videography, art, design, and scientific fields. However, some of these ventures by not just the main players, but hundreds of companies are posing a threat beyond that of what most recognize. I write this because of one such comment on a post of mine today that claimed voice-cloning AI is harmless. This is my side of that comment in full form, and I am going to dive deeper into other aspects of the recently emerged technology.

Voice-Changing AI tools given it spurned this article:

We have all heard the news and the warnings from CISA in regards to the election and deepfakes including vocal deepfakes from the candidates and other known figures, however there is a much more intricate aspect to the risk posed by the Voice Cloning technologies. What you ask?

Fifteen Seconds is what OpenAI is claiming their tool requires to completely mimic a voice, and that they are placing safeguards. These can only be achieved in two ways off the top of my head:

1. Famous and Political / Influential People will be sampled and excluded
2. The entire world needs to submit samples which in all reality with cell phones probably already have but would have to give consent to protect using a stored version of your voice and the world’s voice.?

Why do I say #2 is needed to ensure total security? What is a phone call you could do in fifteen seconds of talking with someone?

A situation that I have written about for months that occurred a couple of weeks ago which I would assume this is how it was achieved. Steps to achieve a hack of a network using this technology and the ease:

1. Call a company’s main line
2. Ask for the Director of IT
3. Hit Record
4. Say that you are looking for a job, you have the wrong number, anything to have him speak for 15 seconds.?
5. Run this voice through the systems that can clone voices, including the yet to be released OpenAI tool.
6. Put your speaker phone on
7. Call back and ask for the target department preferably not YET the IT Department as the director is most likely near enough for the person to know if they are being called by them.
8. Call a Company Employee and you have access to the network of the company using a technology that takes a maximum 15 minute set of calls, some editing, and about 30 seconds to generate any command you would like for the person to do.

Let’s Stay on OpenAI and SORA and other companies versions…for Text to Video

Deepfakes are the obvious problem with these, and it is actually less dangerous in my opinion now than the voice cloning. In institutions where Zoom is still widely implemented, and in remote work this is still a problem but we face the same problems: information provided, orders given, commands via a superior officer all can be mimicked in image, video, and voice and the threat is real.?

Safeguards offer us political and influencer protections but in order to truly protect we are going to have to submit our physical information which is a big ask. Another huge issue is the there are open-source versions capable of being run, and foreign entities not following our regulations. The world needs to prepare, will it be like 1984, the movie Gattaca, or are we just going to let the theft occur as we have since 2017 and WannaCry. Personal Opinion Following: (DON’T READ)All in all it is sad right now the reparations for the hacks are less than the profits, and costs to protect.??