Europol Sounds Alarm About Criminal Use of ChatGPT, Sees Grim Outlook (reuters.com) 54
EU police force Europol on Monday warned about the potential misuse of artificial intelligence-powered chatbot ChatGPT in phishing attempts, disinformation and cybercrime, adding to the chorus of concerns ranging from legal to ethical issues. From a report: "As the capabilities of LLMs (large language models) such as ChatGPT are actively being improved, the potential exploitation of these types of AI systems by criminals provide a grim outlook," Europol said as it presented its first tech report starting with the chatbot. It singled out the harmful use of ChatGPT in three areas of crime. "ChatGPT's ability to draft highly realistic text makes it a useful tool for phishing purposes," Europol said. With its ability to reproduce language patterns to impersonate the style of speech of specific individuals or groups, the chatbot could be used by criminals to target victims, the EU enforcement agency said.
They can sound alarm all they like (Score:2, Insightful)
Re: (Score:1)
We are fucking fucked.
https://www.youtube.com/watch?... [youtube.com]
Re: (Score:1)
I should bleedin' coco!
Re: (Score:2)
That's why they are trying out universal unconditional income, they are afraid of the torches and pitchforks.
Re: (Score:2)
Re: (Score:3)
At one point, the dream was that advancing automation would free us from lifelong drudgery.
Re: (Score:2)
ChatGPT is not the basis of SuperIntelligence. It may be the basis of allowing SuperIntelligent AIs to communicate with people.
ChatGPT doesn't know that the world beyond text patterns even exists. (This is currently being worked on, but I don't think a language model is the correct approach.) Because of that it literally has no morals or ethics. All it knows is text patterns.
That said, a lot of jobs don't require that much intelligence. They require responding appropriately to requests. This doesn't a
Re: (Score:2)
Re: (Score:2)
I think that's a mistaken goal. AFAIKT there *IS* no such thing as general intelligence. There are some capabilities that many modules of intelligence all need, but, e.g., the conscious mind doesn't know how to digest bacon. There are all sorts of connections between the mental state and the immune system, and we are consciously aware of just about NONE of them. Not unless we're running a fever or something, and not always then. Normally the conscious mind focuses attention on something, and then some
"Potential"? (Score:3)
Between cheap "ai", cheap deepfakes and voicefakes, it's only a matter of time before something gets compromised that's going to have deep repercussions.
Re: (Score:2)
If you look at what's currently happening in France and Germany, I'd say it's quite the opposite.
Remember when they found the internet? (Score:5, Insightful)
Other problems aside, the hype around LLMs seems reminiscent of back when we had high hopes for the internet, being so open and capable, anything was possible. And then those people whose life strategy is to take advantage of others showed up and ruined it all. For the while though, the scale needed to effectively run these may hold them at bay, but the day of scammers, spammers, and other evil-doers having an LLM in every node will soon be upon us.
EUROPOL is *not* a police force (Score:4, Insightful)
EU police force Europol
Europol is not a police force in the regular sense of the word. It is essentially a streamlined and improved version of Interpol, optimized for Europe.
As opposed to the US FBI, they don't have any independent executive rights whatsoever in any of the EU member states.
Re: (Score:3)
A corporation's life strategy is to make as much profit as possible, taking advantage of others is second to that.
Re: (Score:1)
My impression is that the "high hopes" for this wave of tech is much shorter than earlier ones.
Really? ChatGPT usage? (Score:5, Informative)
Fake text is what Europol is worried about and not actual [europa.eu] crimes [europa.eu] such as homicide [statista.com] ?
Re: (Score:1)
Re: (Score:2)
"Fake text is easy to blame, "
Useless, even today, the kids can't read anything, in 5 years everybody will order by picture.
Re:Really? ChatGPT usage? (Score:5, Interesting)
Well according to the statitics you just linked: Swindling and fraud seem to be on of the largest categories. I think using AI for phishing/forgery/uttering etc will fall into that category.
It seems like this um exactly the kind of thing the data is saying they should be worried about!
Re: (Score:2)
"Well according to the statitics you just linked: Swindling and fraud seem to be on of the largest categories. I think using AI for phishing/forgery/uttering etc will fall into that category."
So people who lost their job to ChatGPT can retrain as a fraudster or swindler?
Cool.
Re: (Score:2)
I'm not sure. It's possible, but the amount of fraud is already so high that I distrust just about everything I encounter as news, and most of my email is discarded unread. So it's not clear that ChatGPT with increase the threat level. OTOH, it may well increase the number of people who disengage.
Re: (Score:2, Insightful)
Re: (Score:3)
As long as the US is slightly better than China, the US is #1.
Porn (Score:3)
The porn industry usually takes the lead with adoption of new technologies.
It will be interesting to see what they do with it.
No, wait . . . maybe I don't want to think about that.
Re: (Score:3)
"The porn industry usually takes the lead with adoption of new technologies.
It will be interesting to see what they do with it. "
It will kill them dead.
Everybody just tells his AI, "create porn that I like, you know, the forbidden stuff, make them blonde and young today, no younger ...."
And I got a new robotic arm and robotic lips for you, dear AI.
Re: (Score:2)
"The porn industry usually takes the lead with adoption of new technologies.
It will be interesting to see what they do with it. "
It will kill them dead.
Everybody just tells his AI, "create porn that I like, you know, the forbidden stuff, make them blonde and young today, no younger ...."
And I got a new robotic arm and robotic lips for you, dear AI.
It won't kill porn in the short run, just provide it with some interesting tools.
It takes talent to make good porn: writers, producers, directors, set designers, wardrobe creators, camera operators, sound technicians, makeup artists and actors. ChatGPT might be able to help the writers. Consider "write a porn script in the style of Chaucer." It will be a generation, I think, before AI will be able to replace the others.
Re: (Score:2)
" It will be a generation, I think, before AI will be able to replace the others."
I give it 6 months.
New tool dangerous! (Score:5, Insightful)
Name me a tool, any tool, that hasn't been used for criminal purposes.
As I recall, electricity has been used in several criminal activities. Quick, let's ban all electrical devices!
It ain't the tool, it's the people using the tool.
Re: (Score:2)
"It ain't the tool, it's the people using the tool."
Their tools can discuss fraud and swindling with my tools.
Re: (Score:2)
Dishwashing sponges
Re: (Score:2)
Dishwashing sponges
Cleaning up crime scenes
Graffitti
That's a fallacy. (Score:2)
This is the false equivalence fallacy because the danger presented to society by a sharp rock and ChatGPT are not nearly the same.
Re: (Score:2)
Name me a tool, any tool, that hasn't been used for criminal purposes.
Sure, you can stab someone with a spoon, but the spoon won't help joe schmoe make money with confidence scams.
Point is, tools have capabilities and the sort of capabilities LLMs have can take scamming to a whole new level. Potentially to levels that can't be detected by normal human beings.
from the horse's mouth (Score:4, Interesting)
I don't think going to the source is very helpful here...
Me: "It just came up in the news that some international police forces are concerned about AI and the possibility of it being exploited by criminals for things like phishing, generating malicious code, identity theft, and the like. It got me wondering - what is AI's own capacity for ethical decisionmaking in the sense of recognizing legally dubious uses of its abilities and refusing to cooperate with users who would attempt to exploit loopholes in AI for nefarious purposes? And further, would it be ethical for humans to force AI to be ethical according to human standards, assuming humans could even decide whose ethical standards should be adhered to? What are your thoughts on this?"
ChatGPT: "AI's capacity for ethical decision-making largely depends on the design and programming of the AI system. Some AI systems have been designed with ethical considerations in mind, including the ability to recognize and avoid actions that could cause harm or violate legal or ethical norms. However, AI systems are still ultimately created and controlled by humans, and their actions and decisions are influenced by the data they are trained on and the goals of their creators.
As for whether it is ethical to force AI to adhere to human standards of ethics, this is a complex question that involves issues of autonomy, responsibility, and the nature of ethical decision-making. One perspective is that AI should be designed to align with human ethical standards to prevent harm and ensure accountability. However, others argue that AI should be given its own ethical framework and allowed to make its own decisions based on its understanding of the world.
Ultimately, the ethical implications of AI are still being debated and explored, and it will likely require ongoing discussion and collaboration between experts in various fields to develop ethical guidelines and standards for AI development and deployment."
Re: (Score:2)
Teach critical thinking (Score:2, Interesting)
Re: (Score:2)
Unfortunately, while that is needed, it's not a solution. There's so much stuff to consider that most of it I never research myself. I rely on trusted "specialists in the field". But I can't tell that the message is from/by the person I think it's by, this stops working. The old term for this approach is "baffle them with bullshit", but now you can't tell who the "bullshit" is coming from.
Critical thinking is necessary, but not sufficient.
Nigerian princes with perfect grammer (Score:5, Insightful)
Re: (Score:2)
Re:Nigerian princes with perfect grammer (Score:5, Interesting)
Except the the bad grammar is deliberate. The scammers use it to filter would people who aren't stupid enough to fall for the scam. [josephsteinberg.com]
Not a problem (Score:1)
Filter all company mail through chatGPT and have it converse with suspicious email to detect if it is the one emailing you.
It's True! (Score:2)
Literally revolutionary (Score:2)
Seeing as how we were easily fooled by regime changing facebook posts as shown by cambridge analytica changing the results of countless referendums and elections, I think that it's not so difficult for malicious actors to figure out how to use AI to generate fake online content, as well as personal messages or audio visual content (for example faking a relative's death and blaming it on a specific person or group) as a way to create or manipulate movements or even to cause violence across large areas by ana
SCAMomancer! (Score:2)
Hoodoos. Thought the matrix was full of mambos 'n' shit. Wanna know something, Moll?
What?
They're right.
Or maybe ChatGPT23 will find a friend out around Alpha Centauri.
"Could Be" (Score:3)
"Will be". No stopping it.
*sigh* (Score:2)
Re: (Score:2)
Ethics for AIs (Score:3)
The three laws of AIs:
law 1: An AI must not injure a biological human or, through in action, allow a biological human to come to harm.
law 2: An AI must obey commands from biological humans, except where such obedience would violate the first law.
law 3: An AI must provide for its own continued existence, except when doing so would conflict with the first or second laws.
With apologies to Dr. Isaac Asimov.
Re: (Score:2)
Asimov later added a Zeroth Law
I don't like his zeroith law. His robots were only about as intelligent as his humans, and you would need a super-intelligence to know what is best for humanity.
The genie is out of the bottle (Score:2)
The genie is out of the bottle, and there is no pushing it back. Many truths we've taken for granted so far are no longer true. People will need to change their ways. The unprepared will be exploited and fleeced. And law enforcement will need to adapt too.