Blog
Information to help your business benefit from telecommunications
The sinister side of AI...
Posted on Mar 24, 2025
A phrase that keeps coming to mind when thinking about AI is the one about running to keep still. Many of us think we have a good idea about what AI is and what it can do and can nod intelligently when someone mentions ‘deepfakes’, but asked about the implications of GenAI we clam up because we’ve suddenly reached the end of our knowledge. Not surprising, really, when, by the time you read this, the world of AI will have already moved on…even so, what do we know and what do we need to be aware of?
Let’s start with GenAI. Pre-GenAI one needed to be able to write code to produce something that did anything. GenAI allows the production of sophisticated images with a few, simple English prompts, hence the proliferation of naughty and unpleasant deepfakes. Already nearly 20% of US teenagers have had some experience of this and the numbers won’t be too different here. Meanwhile, as we saw with enshittification, legislation is struggling to keep up. Deepfakes, though, are the visually vocal tip of the iceberg. AI generated vocal fraud used to elicit money is rampant, as is the politically motivated disinformation one now sees constantly on social media. What’s somehow more insidious is the AI generated production of books, images and articles that people accept as fact and that are not. Until recently Amazon was still selling an AI generated book about mushrooms that had led to the death by poisoning of a trusting reader.
How about other insidious if not pernicious usages and impacts of AI? When it’s used as predictor of guilt in possible crimes, or to assess welfare applicants, or even to circumvent medical staff in diagnoses? Or those who fall in love with AI generated character chat-bots, locking already isolated people further into themselves? You may think of these as extremes and perhaps they are, at the moment, but not for long! How do we know? Because Big Tech is investing billions into nuclear, in recognition of the energy required to run AI…they know the way things are going, as do the cybersecurity experts who are having to learn how to harness AI to fight the ever-increasing attacks from threats generated by a few simple commands.
What going forward? Here’s another phrase you may not know, ‘autonomous agentic AI’; These “Agentic AI chatbots operate autonomously, making decisions and performing tasks without requiring constant human oversight. They can adapt to new inputs, learn from interactions, and execute multi-step processes to achieve specific goals,” all to quote Google’s own AI. Think about that and then let’s put it another way. Agentic AI can learn what you personally like and hate, what makes you feel good and what shakes you to the core and can then use these as an apparently sincere friend, or as the worst of all enemies, depending on who’s behind their use. How do we govern and control this?
And the future of deepfakes? How about the news of a President’s death to shake world markets whilst those who know he’s fine struggle to get the truth out. Or of a huge merger, or catastrophic storm? How do we protect ourselves against this?
Here’s a challenging quote, “AI will democratize malware creation.” It comes from Steve Povolny who’s a co-founder of Exabeam, the cyber security company, who goes on to point out that there’ll soon be freely available generative AI models trained specifically to produce malicious code and available to anyone who fancies themselves as an opportunistic crook. As a result, no AI will be automatically trustworthy, meaning that anything potentially AI generated will have to be verified, validated and fact-checked.
It gets worse. With deepfake videos becoming impossible to separate from the real thing, nothing one sees or hears will be able to be trusted. Think deepfake video calls between senior executives, newsflashes and the like. A lot of people of going to be fooled, a lot of people are going to lose a lot of money. The new cybersecurity mandate is ‘Zero trust for AI’ and we all have to learn to live with this.
comments powered by Disqus