AI errors can be fatal, which is why it is so important to monitor the safety and ethics of smart systems
Today, December 28, 2021
The creator of Tom Cruise’s popular viral TikTok deepfake video, things aren’t as grim as people say when it comes to the technology that allows him to portray the Hollywood superstar so convincingly.
Miles Fisher essentially became a double of the famous actor and he managed to fool millions of people with his videos on TikTok. He spoke about the possibilities and disadvantages of deepfake technology in an exclusive interview with NBC News.
“As I find myself the unofficial face of this deep fake movement, it’s important to learn and I’m fascinated by this,” Fisher commented. “This is the bleeding edge of technology.”
According to the man, the technology has achieved the most realism that humans have ever seen. Last year, however, he decided to bet his likeness to the star, contacting Belgian visual effects specialist Chris Ume to create viral videos. The creators of the video have really managed to attract attention in a subtle way. Despite the fact that the parodied actor did not respond to NBC News’ request for comment, Cruise also did not express a request to stop the release of such videos.
BBC, December 28, 2021
Amazon was forced to update its voice assistant Alexa after it invited a 10-year-old girl to touch the prongs of a half-inserted fork with a coin.
This proposal came after the request “challenge to do” – “Plug in a phone charger about halfway into a wall outlet, then touch a penny to the exposed prongs,” suggested Alexa. The company-developer Amazon commented on the situation, promising to fix the error as soon as possible.
«We were doing some physical challenges, like laying down and rolling over holding a shoe on your foot, from a [physical education] teacher on YouTube earlier. Bad weather outside. She just wanted another one,» – the girl’s mother Kristin Livdahl commented on the situation.
At the very least, the action suggested by Alexa is unsafe even for an adult, not like a child: metals have the ability to conduct electricity, and plugging them into live electrical outlets can cause electric shock and fire.
“Customer trust is at the centre of everything we do and Alexa is designed to provide accurate, relevant, and helpful information to customers,” said Amazon though.
“As soon as we became aware of this error, we took swift action to fix it.”
Marktechpost, January 1, 2022
IBM researchers are trying to do everything possible to protect AI from hacking.
Machine learning can be very sensitive to enemy attacks. In an attack, information is designed in such a way as to get the wrong answer from the model: the response must be protected from hacks by identifying vulnerabilities, anticipating new methods, and developing reliable models.
IBM researchers are trying to close the gap between developing and deploying AI models by making them more resilient. You need to consider the possibility of unintended troubles, for example, data corruption, and situations where attackers deliberately attack machine learning models.
Researchers are committed to improving the countermeasures of machine learning models by identifying new risks and supporting machine learning approaches to overcoming the crisis.
The article, as well as videos published on the official IBM website, contains a detailed description of the various vulnerabilities and points that attackers usually hit. In addition, the main most effective at the moment schemes for working with such vulnerabilities are proposed and an algorithm of actions is proposed for the most effective elimination of possible problems.
Stay up to date with what is happening! Get a first look at news, noteworthy research and worst attacks on AI delivered right in your inbox.
Deepfakes are posing a real threat to the ethical side of AI How can AI be made more secure and trustworthy? Helpnet Security, December 20, 2021 Artificial intelligence is playing ...
