OpenAI is Being Used by Threat Actors to Carry Out Malicious Operations

The latest report by OpenAI confirms that ChatGPT is being used by threat actors from China and Iran for their operations. Proofpoint reported the first signs of these kinds of activities in April and since then OpenAI has tried to disrupt over 20 million cyber operations which were abusing the use of ChatGPT for different cyber attacks.

OpenAI first outlined SweetSpector which is a Chinese adversary which is targeting Asian governments. SeretSpector also targeted OpenAI directly and sent them phishing emails with malicious ZIP attachments by disguising itself as the email address of OpenAI employees. If opened, these attachments lead to SugarGhost RAT spreading on the victim's computer.

Further investigation found that SweetSpector was using a bunch of ChatGPT accounts to perform scripting and vulnerability analysis. Threat actors asked ChatGPT a lot of requests like asking for information on specific CVE numbers, asking how internet wide scanners are made, asking for ways to exploit infrastructure, asking for help to debug codes and many other requests.

The second group known as CyberAv3ngers is an Iranian Government Islamic Revolutionary Group which is targeting the industrial systems in Western countries. These threat actors asked ChatGPT questions related to Python scripts, Programmable Logic Controllers (PLCs), and developing custom bash. The third group was Storm-0817, which was also Iranian-based. ChatGPT was used by the group to create Instagram scrapers, debug malware and translate LinkedIn profiles into Persian.

OpenAI banned all the accounts that were made by threat actors and shared all of the information to cybersecurity partners. This shows that AI tools can be used to make offensive operations from planning to execution.

Image: DIW-Aigen

Read next: Meta Faces Backlash Over AI-Generated Northern Lights Suggestion
by Arooj Ahmed via Digital Information World

TikTok Accused of Harming Teens’ Mental Health in Leaked Lawsuit

According to a new report by NPR and Kentucky Public Radio, the TikTok team is aware of the harm TikTok causes to teens and young people. This finding was made after 14 US attorney generals took TikTok to court after suing the platform for causing harm to the mental health of children. Even though the papers those attorneys submitted to the court were kept hidden, some reporters got successful in reading the contents of those papers.

The contents of the papers stated that TikTok is affecting the thinking, analyzing, memorizing and conversational skills of children which is leading to negative mental health and anxiety related issues. The suit also talked about a feature which allows parents to limit their children's time spent on TikTok.

But apparently, this feature only reduces the average 1.5 minutes of TikTok usage. TikTok just claims that they are trying to improve the public trust on the platform but in truth, it isn't doing anything. A TikTok representative said that it is very careless that JJ NPR has published some contents of the lawsuit because it is misleading the public as there is no context to them.

Image: DIW-Aigen

Read next:

• Researchers Have Developed a New Algorithm Called RoVi-Aug Which Can be Used to Train Other Robots

• Instagram Usage Jumps Among Teens in 2024, TikTok Popularity Rises

• Study: ChatGPT Matches Terrorist Themes With TRAP-18, Identifies Cybercrimes

by Arooj Ahmed via Digital Information World

Study: ChatGPT Matches Terrorist Themes With TRAP-18, Identifies Cybercrimes

According to a study by Charles Darwin University (CDU) researchers published in the Journal of Language Aggression and Conflict found that ChatGPT can be used to identify cyberterrorist. The researchers used four sample statements by terrorists that were given to ChatGPT. The researchers asked ChatGPT about the main themes and topics in the text, and what messages it finds behind those statements.

ChatGPT was able to find the terrorists behind those statements and found the motivation and purpose of each text. The themes concluded by ChatGPT were opposition to secularism and apostate rulers, criticism of mass immigration, retaliation and self-defense, opposition to multiculturalism, struggle and martyrdom and more. ChatGPT was easily able to identify clues that hinted at violence.

The themes identified by ChatGPT also matched with Terrorist Radicalization Assessment Protocol-18 (TRAP-18), a tool which is able to identify potential terrorists. The lead author of the study said that ChatGPT can easily be used as a tool to identify terrorists and it doesn’t even need any additional training to do so.

Even though many LLMs cannot take the place of human judgment, they can still act as an assistant to offer valuable investigation clues. Even though there are concerns about weaponization of AI tools, they can still be helpful in providing a great map to identify terrorists.

Image: DIW-AIgen

Read next: Researchers Say LLMs Do Not Have Any Ability to Reason After Finding Out That AI Models Cannot Even Solve Math Problems with Some Changes
by Arooj Ahmed via Digital Information World

Instagram Usage Jumps Among Teens in 2024, TikTok Popularity Rises

According to the latest research by Piper Sandler, Instagram is the most used app among teens but TikTok is also popular. The survey of 13,515 US teens shows 30% are using Netflix more than YouTube (27%). TikTok (39%) is the most liked app among teens, while Instagram (32%) is the second most liked social media app.

On the other hand, Instagram is being used by 87% of the teens in Fall 2024. This is a 7% increase in its usage from last year. On the other hand, TikTok usage increased from 74% in Fall 2023 to 79% in Fall 2024. Snapchat is the third most used app by teens this year, but it saw a 3% decrease from last year. In 2023, 74% of teens were using Snapchat but now its usage has come down to 71%.

Pinterest and Facebook are also seeing some rise in usage among teens. Pinterest saw a 6% increase in its usage from last year. 41% of the teens are using Pinterest this year, making it the fourth most used app. Facebook saw a 2% increase with 30% of teens using the app this year. Meta reported that 40 million young adults are using Facebook this year in the USA and Canada every day. It can be due to AI-recommended content too.

Read next: Researchers Say LLMs Do Not Have Any Ability to Reason After Finding Out That AI Models Cannot Even Solve Math Problems with Some Changes
by Arooj Ahmed via Digital Information World

Researchers Have Developed a New Algorithm Called RoVi-Aug Which Can be Used to Train Other Robots

Robots are being used these days to perform various tasks and their algorithm is important for them to perform the actual work. Researchers at UC Berkeley have developed a RoVi-Aug framework which will augment robotic data and will help transferring the data in other robots. Modern machine learning and generative models have generalized their data which is used in almost all of the models. So, the researchers wanted to make something similar to that for the robots that can generalize their data.

The researchers had been trying to generalize robotic data since the start of this year and had been doing various experiments on it. In their previous research, the researchers realized that there are some challenges in generalizing robotic data too. They found that if the robotic data is unevenly distributed, it can become less effective in teaching other robots the same skills.

But the researchers soon found out that a lot of robots have uneven datasets, including the Open-X Embodiment (OXE) dataset which is widely used for training robotics algorithms. This type of imbalance can limit the performance of robots too. To solve this issue, researchers proposed a new algorithm called Mirage which uses a technique called cross-painting to transform unseen robots into source robots. But there are some limitations of this algorithm too.

First of all, it needs exact robot models and cameras, and cannot adjust with camera angles. As an alternative, the researchers presented RoVi-Aug which is flexible and adaptable and can create synthetic images that show robots tasks from different angles.

RoVi-Aug also doesn’t require any extra processing during its deployment and allows changing camera angles from different perspectives. It can help researchers to train other robots because it has precise camera setups which are essential for robot training. RoVi-Aug is also cost-effective and can help other robots improve in learning and training.

Image: DIW-Aigen

Read next: New Study Shows Parents Prefer AI for Child Healthcare Advice, Raising Concerns
by Arooj Ahmed via Digital Information World

Researchers Say LLMs Do Not Have Any Ability to Reason After Finding Out That AI Models Cannot Even Solve Math Problems with Some Changes

A group of AI research scientists at Apple published a paper to find if machine learning can really do reasoning and thinking. They gave the AI models a simple arithmetic equation and asked it to solve it. It was a pretty simple equation which could be solved easily, and LLMs did it too. But once they added some extra useless information, LLMs got confused and got the answer wrong.

Why do most LLMs answer incorrectly with simple information? This is probably because they are trained on simple data and can only give to the point answers. When we throw in a bit of irrelevant information where the actual reasoning is required, they cannot answer correctly.

The researchers say that LLMs are not capable of actual reasoning so they get confused when more clauses are added in an equation. They just try to repeat the reasoning steps they are taught in their training data. This research just shows that LLMs can only repeat what they are taught. They do not personally do anything. They just use their data to answer specific questions.

An OpenAI researcher says that correct answers can be achieved from LLMs with a little bit of prompt engineering. Even though better prompting can work for derivations, it may not work on contextual data. This just shows that LLMs cannot reason on its own so it is best to not use it for academics.

Image: DIW-Aigen

Read next: Consumer Advisory Group Which? Warns Users To Keep Mobile Number Active After Serious Security Issues
by Arooj Ahmed via Digital Information World

New Study Shows Parents Prefer AI for Child Healthcare Advice, Raising Concerns

A new study by the University of Kansas Life Span Institute finds that parents trust AI more than humans when it comes to health of their kids. This is a really shocking reveal and the study also found out that many parents find AI written text credible and trustworthy too. Parents trust AI that they are now seeking healthcare information from AI, instead of human healthcare workers.

ChatGPT and other AI models are also known for creating many errors and giving false information. So it is a concerning thing that many parents are using AI for their children. The lead author of the study, Leslie-Miller, says that this research was done so the researchers can learn about ChatGPT's impact and potential industry concerns. Before AI parents used to search about healthcare stuff for their children online but now they ask ChatGPT about it.

For the study, 116 parents were gathered and they were given text papers related to healthcare concerns in children. Half of the texts were generated by AI, mostly ChatGPT, and the other half were written by experts. The results found out that most parents couldn't distinguish between AI written content and human content. Although they weren't told that there would be two types of texts, most of them still chose AI written texts as the most reliable.

If parents are going to blindly trust AI, it is important that human domain specific expertise on healthcare information should be presented to parents. AI is also dangerous because it has a tendency to hallucinate which means that it can give responses that are very convincing but in reality, they are made up by it. LLMs are also just trained online which means they do not have real world information and experiences. The lead author suggests that parents should look for an AI system that has been integrated into a system with some expertise. Just stay cautious and always double check AI responses.

Image: DIW-Aigen

Read next: Study Shows Many Advanced AI Chatbots Would Rather Give Wrong Answers than Admit They Do Not Know the Answer
by Arooj Ahmed via Digital Information World