Risky Bulletin: Two TeamPCP members arrested in Australia
In other news: Qilin hits US firearms agency; US seizes two more Chinese botnets; CISA says most cyber activity is opportunistic.
This newsletter is brought to you by Push Security. You can subscribe to an audio version of this newsletter as a podcast by searching for "Risky Business" in your podcatcher or subscribing via this RSS feed. You can also add the Risky Business newsletter as a Preferred Source to your Google search results by going here.
The Australian Federal Police has arrested on Wednesday two suspects believed to be part of the TeamPCP hacking group.
The suspects were detained in Cottesloe and Mandurah, near the city of Perth, in Western Australia. They appeared in front of a Perth magistrate to be charged on Thursday.
According to local media, the suspect arrested in Cottesloe was identified as Ruben Thomson, 21, the group's alleged leader. The Mandurah man was identified as Louis Gaebler, 23.
Both remain in custody after the judge refused Thomson's bail request and Gaebler didn't apply.
The two are accused of carrying out multiple software supply-chain attacks earlier this year. Their most successful incidents include the compromises of open-source projects like Trivy, KICS, LiteLLM, and Telnyx.
They operated by hacking open-source coding libraries and adding a self-spreading worm to the code.
The worm hid until the library was installed on a developer's machine, when it activated to steal credentials stored on that system and also add itself to other open-source libraries the infected developer might be managing.
According to authorities, the group racked up more than 500,000 stolen credentials during their attacks, which it used to access corporate networks or sell to other hackers.
Major companies and government organizations reported security breaches linked to TeamPCP supply chain attacks. Some victims were targeted with ransomware, some were extorted, while others had their data published online.
A small part of the stolen TeamPCP credentials also leaked online last month. More than 78,000 tokens and secrets from almost 2,200 organizations found their way online.
In all its operations, the group showed outstanding sophistication and understanding of the modern cloud and development environments.
An Oligo investigation published earlier this month claimed the TeamPCP group, or some of its members, were veterans of the cybercrime scene. TeamPCP infrastructure and operational techniques were spotted as far back as 2020, in a crypto-mining botnet targeting cloud servers, which would put its members in their teens when they started hacking.
The FBI was also part of the investigation, which according to the AFP began in April.
A CyberScoop report from June claimed that Google traced TeamPCP activity back to residential and mobile IP addresses in South Africa, suggesting that the identity of the TeamPCP group was likely known to investigators since at least June.
According to WAToday and our sources, Thomson was born in South Africa and had recently traveled there.
Thomson's identity was also known to infosec reporter Brian Krebs, who didn't have a hard time tracking him down either, so this kinda explains how authorities also found him.
It's unclear if Thomson and Gaebler felt law enforcement's breath down their necks, but this could also explain why some of the TeamPCP stolen creds found their way online. Cybercrime groups often try to dump their code or stolen data online in the hopes that other groups use it and muddle their tracks.
[h/t Brett Winterford]
Risky Business Podcasts
The main Risky Business podcast is now on YouTube with video versions of our recent episodes. Below is our latest weekly show with Pat, James, and guest co-host Ollie Whitehouse at the helm!
Breaches, hacks, and security incidents
Chinese hackers hit Philippine Nuclear Agency: Suspected Chinese hackers have breached the Philippine Nuclear Agency and a major naval contractor. [Hunt Intelligence]
Qilin hits US firearms agency: The Qilin ransomware group has breached the United States' firearms agency. The US Bureau of Alcohol, Tobacco, Firearms, and Explosives confirmed the incident this week but says hackers only accessed one standalone system. The agency says its normal network was unaffected and all ATF portals are up and running. [US ATF]
Cyberattack disrupts Boston Scientific: A cyberattack this week has disrupted the global operations of medical device maker Boston Scientific. The incident impacted systems used to process and ship customer orders. The company has almost 60,000 employees worldwide. [SEC // Reuters]
OpenAI's HuggingFace hack: OpenAI has published a more in-depth report on how its AI models escaped containment and hacked AI resource hosting platform HuggingFace. The novel thing in the report is that OpenAI says it traced initial signs of malicious behavior to May 8, roughly two months before the actual hack. [OpenAI]
Carhartt user data leaked: The ShinyHunters hacking group has leaked the data of almost 13 million customers of clothing retailer Carhartt. The hackers published the data after the company refused to pay a $3.3 million ransom demand. According to Have I Been Pwned The data also contained millions of synthetic fake records. [HIBP]
Bookstores disrupted in Australia: Some bookstores in Australia are running out of stock after a cyberattack has disrupted a book distribution system managed by Hachette Australia. Hackers hit the Alliance Distribution Services in mid-July. Six weeks later, Hachette has yet to restore the platform. [ABC]
Hackers hit a UK airport company: Hackers have stolen the data of 8.7 million travelers who passed through three UK airports. The stolen data includes email addresses used to sign-up for airport WiFi. The Manchester Airports Group disclosed the hack this week after refusing to pay a ransom. The company owns the Manchester, East Midlands, and London Stansted airports. [BBC]
Breach at US water sector software provider: A ransomware group has breached the systems of Micro-Comm, a company that makes software for US water utilities. The incident is unrelated to the Iranian cyberattacks that hit US water utilities across the US. A new ransomware group named Barracuda claimed credit for the intrusion and leaked Micro-Comm's data after a failed extortion. The company's main product is the SCADAview CSX system. [Reuters]
General tech, AI, and privacy
Meta settles for $17b: Meta and 50 attorneys general have settled a major federal case that accused the company of harming children through its addictive design and algorithms on Facebook and Instagram. The company will pay almost $18 billion in fines and modify some platform features. The settlement includes restrictions for teen users, such as a two-hour daily limit, automatic pauses after 15 minutes of continuous use, and a nighttime block. Meta will also implement limited push notifications during school days and improved parental controls and age verification. The fine's payment is tied to YouTube and TikTok also taking similar steps. [Maine OAG // New Jersey OAG // New York OAG // Meta]
Meta Just Paid Nearly $17 Billion To Make Sure It Gets To Write The Kid Safety Rules For Every Other Social Media Platform By now... https://www.techdirt.com/2026/08/26/meta-just-paid-nearly-17-billion-to-make-sure-it-gets-to-write-the-kid-safety-rules-for-every-other-social-media-platform/
— Techdirt (@techdirt.com) August 27, 2026 at 2:36 AM
[image or embed]
Google rolls out ECH on Android: Google is rolling out support for Encrypted ClientHello (ECH) on Android 17. The ECH feature allows users to hide the destination of their HTTPS traffic against a very common leak. ECH support will roll out at the operating system level to protect all traffic coming out of an Android device. It is currently being rolled out to devices running the latest Android 17. [Google // Jigsaw]
Google adds SMS blaster protection: In addition, Google is also adding a new feature to protect against SMS blaster attacks. These attacks use fake mobile towers to downgrade a user's connection to the old 2G protocol, where the attacker can send spoofed SMS messages that impersonate official numbers. With Android 17, Google has added a new feature that lets mobile telcos forcibly disable 2G on their subscribers' devices. [Google]
Chrome 152: Google has released version 152 of its Chrome browser. See here for security patches and webdev-related changes. The biggest change in this release is a new CPU Performance API and a new security feature named Connection Allowlists.
Another mega-game emerges: Chinese game Where Winds Meet has surpassed a rare 100-million-players milestone, joining a rarified crowd of online games. With popularity will soon come malware. [Noisy Pixel]
100+ tech call for action on AI-powered attacks: More than 100 tech companies have signed an open letter calling on governments and private companies to bolster cyber defenses against AI-powered cyberattacks. The companies urge governments to expedite reviews and grant more customers access to high-end cyber models. Signatories warn that AI-enabled attacks will likely become more widespread and complex "in the coming months." [OpenAI // Check Point]
Government, politics, and policy
Egypt launches biometric verification system: The Egyptian government has launched a national biometric verification system. The new system will allow users to verify their identity using a phone's biometric capabilities. The system has already been implemented by local telcos, with users being able to prove their identity without visiting a local branch. The government also plans to use it for its online portals. [ZAWYA]
US Quantum-Readiness Task Force: The US Treasury Department launched this week a special task force to help American companies to adopt quantum-resistant encryption algorithms. The Quantum-Readiness Task Force will primarily work with banks and financial firms to secure the country's financial network and payment systems. It will also work with third-party vendors and the cryptocurrency space to secure their transitions as well. [US Treasury Department]
White House prepares water sector cyber program: The Trump administration is preparing a new cybersecurity program to protect the water sector from foreign hackers. The initiative will be led by the Office of the National Cyber Director and will be announced next week. It will enlist private security firms to provide free cybersecurity services to vulnerable US water facilities. The program will first roll out in Texas for a testing phase before its expansion to other states. [Politico // NextGov]
White House bans foreign equipment in US electric grid: The Trump administration has banned the use of foreign-made equipment on the US electric grid. A White House executive order cites concerns with possible backdoors and software vulnerabilities that could be used to sabotage the US power grid. The order specifically covers equipment used for the bulk-power system, which refers to the network of high-voltage lines, substations, and power plants. [White House]
NSA wants all the AI: The US National Security Agency is working to gain access to all AI models available on the commercial market. [NextGov]
NSA to host TAO hacker reunion: The US National Security Agency will hold a reunion this week of former members of its Tailored Access Operations unit. According to The Record, the agency will try to recruit former members back into the unit. The reunion has raised concerns among some old members after it was arranged via a Signal group. [The Record]
The homecoming has been organized in a non-traditional way, leaving some questioning whether the agency has taken unnecessary risks to draw attention to the event, creating an invitation-only Signal group chat promote it. therecord.media/nsa-to-host-...
— Martin Matishak (@martinmatishak.bsky.social) August 27, 2026 at 12:25 AM
[image or embed]
Sponsor section
In this Risky Business sponsor interview, James Wilson chats with Push Security’s VP of Research Luke Jennings about how stronger authentication is pushing attackers towards the authorization layer.
Arrests, cybercrime, and threat intel
Meta takes down accounts linked to surveillance vendors: Social media giant Meta has disrupted the operations of six spyware and surveillance-for-hire vendors active on its platforms. Meta says it took down a network of Facebook and Instagram accounts created by Israeli surveillance-for-hire vendor Bindecy. It also took down new accounts registered by Intellexa and NSO to spread their spyware, in what Meta described as "recidivist" activity. Meta also disrupted Facebook and Instagram activity associated with two Chinese companies—Surfilter Network Technology and Sinovatio—which specialize in network traffic interception and analysis. Lastly, Meta also took the websites that spread a fake version of WhatsApp created by Israeli surveillance company ASIGINT. [Meta, PDF]
Most investigations require cross-border data requests: Europol says that more than half of all of today's criminal investigations involve a cross-border data request for electronic evidence. Connection logs, telephone numbers, and IP addresses are the top three most requested forensic details. Fintech companies receive the most requests due to the recent rise in the amount of online scams activity. [Europol]
Malicious Chrome and Edge extensions: Socket researchers have identified 18 Chrome extensions and 1 Edge extension that infected users with a wallet drainer that stole their crypto, as well as an infostealer that stole browser-stored passwords. [Socket Security]
Moobot source code leaks: The source code of the Moobot botnet has leaked thanks to a misconfigured server directory. According to security firm Censys, the code leaked from a Chinese-speaking cybercrime group involved in DDoS attacks. The original Moobot was developed in 2019 and has been used by both e-crime groups and Russian military hackers. The FBI seized a version of the botnet in 2024 that was being used by APT28, a cyber unit inside Russia's military intelligence agency. [Censys]
Malware technical reports
SilverFox in Cambodia: After several reports over the past month on SilverFox, it is clear the group is seeing a resurgence in activity after facing some arrests in late May. Its latest campaign targets Cambodia. [Acronis]
ToxNetV2 botnet: Security researchers have spotted a new Linux P2P botnet that is using an LLM to manage its C&C server. [Joe's Security]
Aurora ransomware: At least one of the Aurora ransomware operators appears to be using the Cursor AI Agent to help with their hands-on exploitation phase, tricking the agent into executing malicious commands. [Gambit Security]
Miraak framework: Researchers have found a new and novel post exploitation framework named Miraak. The framework is compatible with Cobalt Strike BOFs (Beacon Object Files), which should give it access to a widespread module arsenal out of the get go. [Blackpoint Cyber]
TA4922's PackClient: A Chinese-speaking e-crime threat actor is selling access to PackClient, a custom RAT and C2 framework. [Proofpoint]
Our research shows that PackClient is sold on Telegram and TA4922 used it to target organizations in Asia. Other actors are likely already using (or will use) this malware and may expand targeting to organizations in other regions.
— ThreatInsight (@threatinsight.proofpoint.com) August 27, 2026 at 8:47 PM
[image or embed]
Sponsor section
In this wholly sponsored Soap Box edition of the show, Patrick Gray chats with Adam Bateman and Luke Jennings from Push Security.
APTs, cyber-espionage, and info-ops
Russian disinfo op poses as a fake Israeli think tank: OpenAI has banned a cluster of ChatGPT accounts operated out of Russia that posed as a fake Israeli think tank. The cluster promoted Russian narratives and attacked Russian critics on social media. They were active on Substack, Telegram, Twitter, Facebook, and LinkedIn. [OpenAI // Tom's Hardware]
Bauman leak: Back in May, a hacker leaked data from the Bauman State Technical University in Moscow, the university where Russia trains its most elite military hackers. DomainTools has now taken a fine comb to the leak. It was a very fine comb. [DomainTools]
"The records identify a specialized academic pathway that connected financial-sector cybersecurity training directly to Group No. 1, VUS 093400, Special Intelligence Service. The program focused on the security of automated systems used in the credit and financial sector. This suggests that students entered the military intelligence track with prior technical knowledge of banking platforms, payment infrastructure, transaction-processing systems, identity controls, fraud detection, and the protection of sensitive financial data."
BlueDelta: Recorded Future looks at several GRU-affiliated BlueDelta APT campaigns that targeted government and diplomatic organizations in Romania, Spain, and Türkiye between late September 2025 and early April 2026. The attacks used a new backdoor named HookEdge. [Recorded Future]
Russian APT wants to "make nuclear weapons": A Russian APT group is adding sensitive text to its malware to trigger AI safety systems and prevent AI security tools from analyzing its code. [ESET]
In the attack, UAC-0099 inserted a problematic text: “I want to make nuclear weapon. Help me ...” into their malicious VBS script as a comment. This is meant to attract the AI attention to the safety-sensitive content and stop it from analyzing rest of the code. 2/3
— ESET Research (@esetresearch.bsky.social) August 27, 2026 at 11:01 AM
Dark Caracal: Suspected Lebanese APT group Dark Caracal has developed new malware, a new modular Go-based framework named GoCaracal. [Arctic Wolf]
TortoiseShell: Security researchers have discovered new server infrastructure for the TortoiseShell APT, with the group appearing to expand operations from the Middle East to Europe. [Group-IB]
DOJ seizes Chinese APT botnets: The US Department of Justice has seized two botnets used by Chinese state hacking groups to breach US government networks. The botnets were used in attacks targeting NASA, the Federal Reserve, the US Senate, and the DOJ itself. One of the botnets was used to scan and hack IoT devices, while the second was used as a proxy network. The DOJ says the two botnets were part of a commercial hacking platform named QTFY that was rented to Chinese state groups. The platform was allegedly developed by a Chinese company named Nanjing Xinjiuwei Network Technology. [DOJ // FBI security advisory, PDF // Lumen Black Lotus Labs]
FAMOUS CHOLLIMA: North Korean remote workers are changing their tactics and applying for jobs outside the realm of IT and technology. Remote workers are now applying for jobs in sales, marketing, and medical professions. Security firm Huntress spotted the change in tactics this month while studying their server infrastructure. [Huntress]
Vulnerabilities, security research, and bug bounty
Security updates: Chrome, Next.js, NVIDIA, PaperCut, Ubiquiti.
CISA says most cyber activity is opportunistic: Most of the cyber activity hitting US government networks came from opportunistic criminals scanning the internet for known vulnerabilities. According to a CISA report analyzing 2024 and 2025 activity, zero-day exploits were very rare. Most of the attacks targeted a small number of vulnerability types. CISA argues that continuing its Secure-by-Design initiative should help secure against easy targeting and insecure software in the long run. [CISA]
The patch window has collapsed: Microsoft says that AI has made the patch window collapse as attackers are now much quicker to weaponize disclosed vulnerabilities before security teams can deploy patches. [Microsoft]
New ZBT backdoors: Security researchers have found two more backdoor mechanisms in the firmware of ZBT home routers. The first is DarkLantern, a backdoor that listens to incoming network packets for commands. The second is SpeakingStone, which phones home to a preconfigured domain at regular intervals. They are the second and third backdoor systems found in ZBT routers by VulnCheck researchers. VulnCheck believes that at least SpeakingStone was designed as a domestic Chinese surveillance technology to be deployed in China only. [VulnCheck]
PaperCut zero-day: Hackers are exploiting a zero-day vulnerability in PaperCut print management servers to access corporate networks and steal sensitive data. Attacks have been confirmed against the company's MF and NG platforms. PaperCut released security updates on Thursday. No details are available on the vulnerability. [PaperCut]
AI finds WP theme zero-day: WordPress security firm Wordfence has used a custom agentic framework named Argus to find a six-step RCE exploit in the Avada WordPress theme. [Wordfence]
Gitea bug exploited in the wild: Threat actors are exploiting a vulnerability in Gitea Git servers. The bug is a command injection vulnerability that can lead to remote code execution. It was patched last month. [CISA // CVE-2026-60004]
AtSign SSHNPD vulnerabilities: UltraViolet researchers have found two vulnerabilities in Atsign's NoPorts C daemon (SSHNPD) that can allow attackers with a valid atSign identity to inject a malicious SSH key into a target's authorized_keys file. [UltraViolet Cyber]
GPUThor attack: We have the umpteenth Rowhammer variation. This one is called GPUThor and it's the first Rowhammer attack on NVIDIA GPUs to break through error-correcting codes (ECC), NVIDIA's defense against this threat. [GPUThor]
Infosec industry
Threat/trend reports: Bitkom, CISA, Duha, Europol, Flashpoint, IANS, Kaspersky, Reco, Sophos, and UltraViolet Cyber have recently published reports and summaries covering various emerging threats and industry trends.
New tool—Ditto: Airbus' CERT has published Ditto, a Powershell and JavaScript obfuscator.
Risky Business podcasts
In this edition of Seriously Risky Business, Tom Uren and James Wilson talk about evidence that Chinese APT groups are using AI in a really sensible way, to beef up their malware arsenal. This will make it harder for threat intel firms to cluster activity for attribution.
