Warum ChatGPT für uns gefährlich werden kann

Okay, admittedly, AI can be quite funny, like with the famous AI pope's photo. But Italy basically prohibits chat GPT in its own country and Volker Wissing wants to regulate chat GPT and co. Because it can also be used as a digital weapon. I think many of you, including me, have not thought about the fact that chat GPT and co. can also be used for hacker attacks. Also things that are now being considered by the federal minister, such as Volker Wissing. And maybe you remember, the Bundestag has also already been victim of digital attacks. For days, the Bundestag was on the net in 2015, i.e. espionage software on computers by politicians such as Angela Merkel included. And today, for example, Russia has developed systems with which it wants to have the opportunity to let trains derail. And AI systems could also be used in Germany for such cases. In this video, we therefore take a look at the announcement by the government or Volker Wissing, the Federal Minister of Digital Affairs, whether chat GPT must be banned or whether it should be regulated. What is the plan there? And with that, welcome back to a new video. And after this somewhat long intro, let's take a look at the topic in three steps. First, what criminal dangers do chat GPT and co. pose? Second, with which real hacker attacks and digital weapons do we have to expect? And third, how does the government possibly imagine such regulation or such a ban? By the way, how do you see it? Does Volker Wissing have to ban chat GPT in Germany? Yes, no, or are you not sure? Your opinion in the comments. We all know that chat GPT can write bachelor's theses or that the whole website is encoded. But 51% of leading IT experts from North America, Australia and the UK actually expect that there will be a successful cyber attack by the end of 2023, which can be traced back to chat GPT. By the way, a survey conducted by Blackberry. And that's right, Blackberry, which once produced mobile phones and smartphones. In the meantime, the company has fully focused on online security and has written it down. And 71% of the surveyed experts also assume that states are already making use of chat GPT for malicious purposes. And among the top 5 fears of how criminals could use it, there are things like First of all, to increase the roughness of threats and attacks, 46% said that. Or also to spread misinformation, 49% said that. Or also to help hackers to create more credible and more serious phishing emails. Furthermore, the chief technologist or director of Blackberry said, quote, In fact, both cybercriminals and cyber defenders are actively investigating how they can use chat GPT to strengthen and improve their intended results. And they will continue to do so. Time will tell which side is ultimately more successful. And Blackberry is not wrong. Because the European Police Authority, Europol, also looks at the whole thing in more detail. At the beginning of 2023, investigators of the authority dealt more closely with the chatbot. And investigators of the authority published an official report, in which they warned. The capabilities of chat GPT are suitable for a number of potential misuse cases in the area of terrorism, propaganda and disinformation. So the model can be used to collect more information that could facilitate terrorist activities. Such as terrorism financing or anonymous file exchange. And it has been known for a long time that Germany was not quite so well positioned in digital cyber defense. The report and research by Böhmermann are welcome there. Or here, too, an excerpt from a meeting in the Bundestag. I don't know who of you was in the LKA last time, in the Cybercrime Unit. These are not conditions under which you can now pull a lot of people out of the university as defendants. That is not attractive. So here, too, create incentives, give the means, so that the prosecution can do what it is actually there for. Investigate. Investigate perpetrators. So, and chat GPT is a new building block that the authorities now also have on the screen. Also a reason, by the way, why Federal Minister of Digital Affairs Volker Wissing now demands a regulation. But first, let's get to the dangers. Let's take a look at how such digital attacks, for which AI is also used and can be used, really look like in reality. And maybe you remember that for some time now I have been cooperating with SPIEGEL as an advertising partner. And we have prepared some cool things for you. Like now, for example. Because SPIEGEL, together with a few others, has an extremely exciting investigative researcher at the start, the so-called Vulcanfiles. Because a whistleblower has revealed to SPIEGEL internal plans of a Russian company, which builds digital weapons for Russia's intelligence services and the military, and thus enables them to carry out digital attacks. I also spoke to investigative journalist Alex, who also hosts the podcast Putin's War on the Internet for SPIEGEL. I met her in a Berlin hotel and I wanted to know from her why the accusations are so blatant and why cyber security affects us all. I immediately asked if something could happen to me, because someone could potentially bring me around the corner afterwards. And they said, who knows, maybe. So my colleague Hannes Munzinger, who belongs to Papertrail Media, who works for SPIEGEL, received a message shortly after the beginning of the Ukrainian war last year. The message was anonymous. You didn't know who sent it. Hundreds of documents were attached to the message. Documents that belong to a Russian company. The company is called Vulkan. And on the paper they are just a small IT company, or not so small, but an IT company that produces software. And this anonymous source, which Hannes sent these documents, claimed that this, at first glance, harmless company actually produces things for the Russian intelligence service and these things are dangerous. And out of anger about the Ukrainian war, this person, we don't know if it's a woman or a man, sent these documents to Hannes. And we came across more and more things that are really frightening and also potentially dangerous for many people. For example, it was shown to me that Vulkan produces software for the Russian intelligence services and, among other things, software that is supposed to be able to decontaminate trains or set the air traffic alarm. And you have to say, after what we found out, it is not excluded that something like this could happen in Germany. That at some point someone will cut off our electricity. Simply because the Russian leadership wants it. And to realize that it is technically possible and that Russia has already done it and could do it again, that scared me a lot. Russia is not bombing us, but Russia is putting itself in a digital position. And that's how we have to see it. It's a way of waging war. That's how Russia sees it. And if you now want to dive deeper into the Vulkan files and the articles, the podcasts and everything else, then you can do that now. Everyone who is 30 years older will get a free month of Spiegel Plus And everyone who is under 30 will get it for 5.99 euros for the first 6 months. And with that you can not only hear the latest Vulkan files podcast episode a week earlier than everyone else, which is pretty nice, no. You can then of course also read all the Spiegel Plus texts with the Spiegel Plus subscription. You can watch the digital edition, the Plus newsletter and get a lot more. And I think such a cooperation cannot be better. So, as I said, you can find the offer in the pinned comment or in the description. But now let's get to the third one. A ban on chat GPT. Is that in front of us in Germany? So you see, the dangers are real. And while in Italy, chat GPT pulls the Italian plug, so to speak, and forbids the company from using Italian user data, Volker Wissing, the Federal Digital Minister, thinks that this was going too far. That he doesn't want that for Germany. Volker Wissing is against such a ban, as he himself explained in an interview. We now have to react smartly and reasonably regulate artificial intelligence before it's too late. This must not take years again. He also means, for example, that one must ensure by law, that this new technology can only be used if it adheres to European values such as democracy, transparency and neutrality. AI systems must not be manipulated, they must support us. And he does not want to leave totalitarian powers, at least chat GPT. Quote, we must not leave him the field. We need European AI investments that are based on a reliable legal basis. What you should know in this context is that the EU is already working on such a regulation of chat GPT & Co. Namely with the AI Act or AI Act, whatever you want to call it. This law came into force at the end of April 2023, but not yet in force. But the EU plans to subdivide different AI systems into different risk categories. AIs that can, for example, design texts, quote, of which one could falsely assume that they were written by a person, quote, for example, fall into the high-risk category. There should be four categories. Minimal, limited, high and unacceptable risk. Yes, and which unknown AI can do that right now? Chat GPT. But what happens with these different risk categories? AI for mass surveillance or manipulation of people would be unacceptable, i.e. category unacceptable and therefore categorically prohibited. But how exactly penalties or rules could look like for the individual categories, that is not really clear yet. But it is clear that the higher the classification, the more restrictions, more rules, more penalties. This could either be due to the fact that providers like chat GPT would probably no longer offer the product in Germany or that the manufacturers would adapt their products themselves, which would result in them falling into a lower category. By the way, WBS-Kanzlei has already made a more detailed video about this, which I have of course linked to you up here in the i. But as I said, your opinion on the subject now in the comments below and test out the free Spiegel Plus subscription or the discounted Spiegel Plus subscription for you. All in the description. And thank you for watching. And here's another video of mine if you want to watch it.