AI-generated spam will soon be flooding your inbox — and it’ll be personalized to be especially persuasive.


Every day, e-mail inboxes are choked with messages from Nigerian aristocrats, drug lords, and promoters of dodgy investments. Improvements in spam filtering seem to inspire new techniques to bypass protections.

Now, the arms race between spam blockers and spammers is about to intensify with the emergence of a new weapon: generative artificial intelligence. With the latest developments in AI ChatGPT has become popularSpammers may have new tools to evade filters, grab people’s attention, and convince them to click, buy, or give up personal information.

As director of the Laboratory for Advancing Human and Machine Reasoning at the University of South Florida, I do research The intersection of artificial intelligence, natural language processing and human reasoning. I studied how AI can learn people’s individual preferences, beliefs and personality traits.

This can be used to better understand how to interact with people, help them learn, or provide useful advice. But that means you have to rely on smart spam that knows your weak spots and can exploit them against you.

Spam, spam, spam

So what is spam?

Spam is defined as unsolicited commercial emails. Sent by an unknown entity.. The term is sometimes applied to text messages, direct messages on social networks and Fake reviews on products. Spammers want to get you to take action: buy something, click on phishing links, install malware, or change views.

Spam is profitable. One email blast can earn $1,000. in a few hours, they charge spammers just a few dollars – not including initial setup. It can create an online drug spam campaign. Around $7,000 per day.

Legitimate advertisers want to get you to take action—buy their products, take surveys, sign up for newsletters—but a salesperson’s email may link to an established company website and contain an unsubscribe option. Federal regulationsmay not be spam email.

Spammers do not have access to mailing lists that users subscribe to. Instead, spammers like the “Prince of Nigeria” fraud, a Nigerian prince has promised to reward him handsomely, saying he wants your help in unlocking a virtual currency. Intelligent digital natives will immediately reject such requests, but the stupidity of the question Of course, he may choose for naivete or old age.Check out the most likely to fall into scams.

Advances in AI, however, mean that spammers may no longer rely on such hit-or-miss approaches. AI allows them to target individuals and make their messages persuasive based on easily accessible sources like social media posts.

The future of spam

You may have heard about developments such as large language models. Discussion GPT. The task these generative LLMs perform is deceptively simple: given a text sequence, predict which token—think of it as a word element—comes next. Then predict which sign will come. And so on, repeatedly.

Somehow, just training on that task, complete with enough writing on a large enough LLM, seems to be enough to develop the ability to run these models surprisingly well. Many other functions.

Many ways to use the technology have emerged, the technology shows the ability to adapt and learn quickly with individuals. For example, LLMs can write entire emails in your style with just a few examples of how to write. And there is the famous example – which is now more than ten years old – of Target Knowing that a client is pregnant before her father knows.

Spammers and So are marketers It is useful to be able to predict more about individuals with less information. Armed with your LinkedIn page, a few posts and a profile picture or two, LL.M.-armed spammers can make a reasonably accurate guess about your political leanings, marital status, or life priorities.

Our research shows that LLMs can be used to predict with some accuracy which word an individual will say next. Far superior to other AI approachesIn a word-generation task called Translation efficiency function. We have also shown that LLMs can take certain types of questions from reasoning tests Predict how people will respond to the question. This suggests that LLMs have some knowledge of what normal human reasoning is like.

If spammers get past the first filter and get you to read an email, click a link or engage in a conversation; Their ability to apply custom persuasion is greatly enhanced. Here again, LLMs can change the game. Early results indicate that LLMs can use it to argue persuasively on topics ranging from Politics to the Public health policy.

Good for Gander

But AI doesn’t favor one side or the other. Spam filters should benefit from advances in AI, which will allow them to create new barriers to unwanted emails.

Spammers often try to trick filters. Special characters, misspelled words or hidden textRelying on the human tendency to forgive small textual anomalies – eg “c1îck h.ere n0w”. But as AI gets better at understanding spam, filters can become better at identifying and blocking unwanted spam — and perhaps pass through desirable spam, like marketing emails you’ve clearly signed up for. Imagine a filter that predicts whether you want to read an email before you read it.

Despite growing concern about AI — as evidenced by Tesla, SpaceX and Twitter CEO Elon Musk, Apple founder Steve Wozniak and other tech leaders Calling for a pause In the development of AI – many good things can come with the development of the technology. AI It can help us understand it. How weaknesses in human thinking can be exploited by bad actors and how to create ways to counter malicious behavior.

All new technologies are exciting and potentially dangerous. The difference lies in who creates and controls the tools and how they use them.

This article has been updated to reflect the father of the teenager who learned from Target that his daughter is pregnant.

The Conversation

John LicatoAssistant Professor of Computer Science and Director of the AMHR Laboratory University of South Florida

This article was reprinted from The conversation Under a Creative Commons license. Read Original text.

We offer you some site tools and assistance to get the best result in daily life by taking advantage of simple experiences