Risky Bulletin: BGP hijack delivers malicious Virtualizor updates

0:03A BGP hijack delivered malicious virtualiser updates. The White House launches Project Watershed 250. Indian authorities take down a Telegram doxing bot. And composer packages deliver iOS badness. This is the Risky Bulletin, prepared by Katalin Kimpanu and read by me, Claire Aird. Today is the 2nd of September, and this podcast episode is brought to you by Dropzone AI. In today's top story, hackers staged a BGP hijack attack against the software company Softaculous and delivered malicious updates to its customers. Softaculous makes Virtualiser, which is used by web hosting companies to manage virtual private servers.

0:48The Virtualiser team says the hacker performed the BGP hijack, obtained a TLS certificate in its name and hosted a clone website that delivered the malicious updates. The BGP hijack was live for almost 33 hours over the weekend. The number of organisations that downloaded the malicious updates remains unknown. In other news, the hacker attempted to steal$75 million in crypto tokens from the Tectonic DeFi platform, but only got away with six. The attacker exploited a bug in the Cosmos blockchain platform that Tectonic runs on. The hacker initially made off with$74 million, but Tectonic clawed back$68 million in the hours following the attack.

1:31Last week, the same Cosmos bug was also exploited to steal almost$3 million from six other platforms. Cosmos says it fixed the bug last week, but has now halted all activity. AI model evaluation organisation Meta has disclosed two security incidents that took place earlier this year. In March, a hacker stole an API key and consumed credits worth$600 ,000. Then in May, a second group probed META's public infrastructure but failed to steal data. The company says the incidents impacted back-end infrastructure but no sensitive data was accessed. META is currently helping both Anthropic and OpenAI investigate security incidents where their cyber models escaped test environments and hacked real companies.

2:20More than 100 ,000 files have been found leaking from the server infrastructure of Andrew Tate's multimedia business. The leaked files contained staff chats, videos from private events and 76 paid course videos from Tate's War Room program. The courses taught members how to approach women, bribe police and government officials and spy on and exploit rich and powerful people. The files have been made available for download via the DDoS secrets portal. Anthropic has paused external cyber evaluations of its pre-release models. The company cited recent incidents of models escaping their test environments as the reason for the halt.

3:02Anthropic also paused reinforcement learning for some models while the company strengthens its monitoring. OpenAI also paused reinforcement learning after its agents hacked Hugging Face last month. The White House has launched a pilot program to protect the water sector from foreign hackers. The Project Watershed 250 program will connect water utility providers with cyber defence resources at no cost. The program will run for a six-month test phase in Texas before a possible expansion to other states. The White House will require all government agencies to switch to the login.gov platform as the default user authentication method for public-facing government websites.

3:45According to a memo from the Office of Management and Budget, agencies have two years to make the switch. The US launched login.gov in 2017. The US House of Representatives has passed a bill that will allow telcos and tech companies to notify users after their data has been subpoenaed, even when a secrecy order is in place. Under the NDO Fairness Act, non-disclosure orders preventing the companies from notifying users will be limited to 90 days. The bill passed unanimously in the US House and is headed for a Senate vote. Indian authorities have taken down a telegram bot that allowed threat actors to buy detailed personal information on any Indian citizen for as little as 6 US cents.

4:33According to security firm Cideres, the platform used legitimate credentials to access the data from at least three KYC providers blending its queries inside their traffic. Cideres has linked the platform to a threat actor named Mr X is Back. Five Venezuelan nationals have pleaded guilty to hacking ATMs across the state of Kansas. The group installed malware on the device to force them to dispense their cash in what's known as an ATM jackpotting attack. The suspects were detained last December as part of a major bust of Venezuelan ATM jackpotting gangs. A newly formed pro-Ukrainian hacking group is targeting Russian companies with ransomware.

5:17According to Russian security firm F6, the group has hit at least seven victims and has demanded ransoms worth millions of US dollars. The group uses its own custom ransomware and appears to include members from previous older Ukrainian hacker groups. A hacking group targeted more than 150 employees across 10 companies in a coordinated social engineering campaign. The hackers used external Microsoft Teams accounts to contact employees while posing as help desk staff. The attacks sought to trick employees into installing malware or legitimate remote management tools that could be used for future access.

5:57The campaigns took place between January and April. Palo Alto Networks tracks this threat actor under the name Spring Ring. A cluster of 13 malicious composer packages are injecting malicious JavaScript code on websites built by Vietnamese developers. The code shows website visitors unwanted ads, redirects them to gambling sites and attempts to run iOS exploits that install spyware on their smartphones. The spyware can steal passwords and cryptocurrency wallet seeds. Apple patched the exploit last November with iOS 26.1. Security researcher Nightmare Eclipse has published two more zero-day exploits on GitHub.

6:39The first gives full read-write access to the Windows operating system via NVIDIA GPU drivers. The second grants attackers system access to the operating system via the Avast antivirus. The two are named Green Section and Pretty Prague, respectively. This is the researchers' third zero-day in as many days after they also disclosed a vulnerability in Kaspersky's EDR. Threat actors are exploiting a recently patched authentication bypass vulnerability in JFrog Artifactory Artifact Repository Management Service. According to security firm Watchtower Labs, exploitation began three days after a patch was released last week.

7:22It's a different vulnerability to the one OpenAI agents use to hack Hugging Face in July. Hackers are exploiting an SQL injection vulnerability in Sangoma switchbox VoIP telephony servers. Attacks were spotted last week after patches were released in July. Exploitation does not require authentication. The GeoNetwork project released security updates on Monday to patch four vulnerabilities that can be used to take over geospatial servers. The project's software is a web interface for viewing custom geospatial maps and was developed by the UN's Food and Agriculture Organisation. Security firm ETHIAC has identified vulnerable servers in 39 countries.

8:07German company Cryptware has released security updates to patch nine vulnerabilities in its CryptoPro software. The software is typically used in ATMs and other Windows-based embedded devices. The bugs can be exploited to bypass disk encryption and pre-boot authentication and to read clear text data from the devices. Major ATM vendors like Diebold Nixdorf have already rolled out patches. And finally, a team of South Korean academics has developed a$7 phone case that can detect the presence of hidden cameras in a room. The phone uses a special app to illuminate objects from different angles and analyse reflections to spot hidden lenses.

8:48The device recorded a 94 % accuracy during testing. And that is all for this podcast edition. Today's show was brought to you by our sponsor, Dropzone AI. Find them at dropzone.ai. Thanks to your company.