Security Blog

AI – the good, the bad and the downright scary

Artificial intelligence brainArtificial intelligence (AI) is an advanced technology that works quite similar to the human brain, as AI can register, learn, understand, and act like us. AI can drive innovation in a lot of areas, and a lot of companies are now exploring ways to use and incorporate the technology into their services and processes.

Mentioning artificial intelligence usually evokes one of two responses – excitement or fear. And, it is difficult to imagine how AI will develop – where the most loaded question is what will happen when we build machines that are more intelligent than we are?

The potential of AI is undeniable. The technology can improve growth, be used across a lot of different industries, has the potential to cure diseases, and will probably perform our jobs better than we do in many cases. As with any new development, there are risks, and information security in relation to AI needs to be addressed. And, rather urgently with the extremely fast development of AI in areas raise concern around personal integrity and security.

Deepfakes and Deep Voice

Some of you might have heard of deepfakes, the phenomenon of using machine learning to essentially paste the face of one individual onto another in videos. It started with porn where some of the women’s faces were swapped with faces of celebrities like Gal Gadot and Scarlett Johansson. The technology has also been used more light-heartedly to face-swap Nicholas Cage into movies like James Bond and Indiana Jones. Now, online platforms like Reddit have banned AI assisted face swapping videos and images, but as the technology becomes more sophisticated, can you guarantee the authenticity of an image?

This is starting to sound like an episode of Black Mirror, right? Well it gets scarier, or cooler, depending on your thoughts on AI.

Recently, we have also seen the emergence of ‘Deep voice’ – software that with the use of an AI algorithm can clone a voice. Using just 3.7 seconds of audio, the Chinese company Baidu can clone a voice, and even alter the accent, speech and tone. Motherboard is where you want to go if you want to hear some examples. Although being an innovative use of technology, there is no denying that the software enables you to potentially fabricate interviews, news segments, announcements etc. with the use of a short audio clip. And, how will this software and the existence of doppelganger voices impact voice recognition programs – will they become obsolete?

The need for a discussion on AI and information security

AI can use personal data – such as voices and pictures, in quite malicious ways, which highlights the importance of information security and legislations, processes and routines to protect information. Companies using and developing AI should have security measures in place, and take steps to ensure that the potential of the technology does not mean a security compromise.

  • 24 Solutions AB
  • Smedjegatan 2C
  • SE-13154 Nacka, Sweden
  • +46 (0)8 535 24 100
  • info@24solutions.com