"Google's Naptime Framework to Boost Vulnerability Research with AI"

"Google's Naptime Framework to Boost Vulnerability Research with AI"

Researchers at Google's Project Zero introduced "Naptime," a framework to allow Large Language Models (LLMs) to perform vulnerability research. Naptime, launched in mid-2023, seeks to improve vulnerability discovery approaches, focusing on automating variant analysis. The Naptime framework lets LLMs conduct vulnerability research like human security experts, mimicking the iterative and hypothesis-driven approach.

Submitted by Gregory Rigby on

"Meta's Virtual Reality Headset Vulnerable to Ransomware Attacks: Researcher"

"Meta's Virtual Reality Headset Vulnerable to Ransomware Attacks: Researcher"

Researcher Harish Santhanalakshmi Ganesan demonstrated the delivery of malware to Meta's Quest 3 headset. He took on claims that it is almost impossible to install malware on Quest 3 VR, and did it without enabling developer mode. Googling led him to a method to install "CovidLock" ransomware on his headset. The ransomware targets Android devices, pretending to be a COVID-19 tracker app and gaining additional permissions to the point where it can lock users out and display a ransom note.

Submitted by Gregory Rigby on

"Indonesia Says a Cyberattack Has Compromised Its Data Center but It Won’t Pay the $8 Million Ransom"

"Indonesia Says a Cyberattack Has Compromised Its Data Center but It Won’t Pay the $8 Million Ransom"

Indonesia’s national data center has recently been compromised by a hacking group asking for a $8 million ransom that the government won’t pay.  Samuel Abrijani Pangerapan, the director general of informatics applications with the Communications and Informatics Ministry, said that the cyberattack has disrupted the services of more than 200 government agencies at both the national and regional levels since June 20.

Submitted by Adam Ekwall on

"Neiman Marcus Data Breach Disclosed as Hacker Offers to Sell Stolen Information"

"Neiman Marcus Data Breach Disclosed as Hacker Offers to Sell Stolen Information"

High-end department store Neiman Marcus recently disclosed a data breach shortly before a hacker offered to sell information belonging to the company's customers.  The company said a database platform storing personal information was compromised between April and May 2024.  The data breach was detected in May. The company noted that an investigation showed that the hacker had gained access to information such as names, contact data, dates of birth, and Neiman Marcus or Bergdorf Goodman gift card numbers.

Submitted by Adam Ekwall on

"Chrome 126 Update Patches Memory Safety Bugs"

"Chrome 126 Update Patches Memory Safety Bugs"

Google recently announced a new Chrome security update that addresses four high-severity memory safety vulnerabilities reported by external researchers.  Google noted that the four issues tracked as CVE-2024-6290 to CVE-2024-6293 are use-after-free bugs impacting the Dawn and Swiftshader components of the popular browser.  The latest Chrome iteration is now rolling out to users as version 126.0.6478.126 for Linux and as versions 126.0.6478.126/127 for Windows and macOS.

Submitted by Adam Ekwall on

"Hackers Steal Over $2 Million in Cryptocurrency From CoinStats Wallets"

"Hackers Steal Over $2 Million in Cryptocurrency From CoinStats Wallets"

Cryptocurrency portfolio manager CoinStats recently resumed activity after hackers drained over $2 million in virtual assets from 1,590 hosted wallets.  The incident occurred on Friday and prompted CoinStats to shut down its application to mitigate the attack.  Because CoinStats asks for read-only access to connected wallets, only some of those created directly within CoinStats were affected by the hack.  CoinStats noted that none of the connected wallets and CEXes were impacted.

Submitted by Adam Ekwall on

"Critical RCE Vulnerability Discovered in Ollama AI Infrastructure Tool"

"Critical RCE Vulnerability Discovered in Ollama AI Infrastructure Tool"

A now-patched security flaw impacting the Ollama open source Artificial Intelligence (AI) infrastructure platform could have enabled Remote Code Execution (RCE). Ollama is a service used to package, deploy, and run Large Language Models (LLMs) locally on Windows, Linux, and macOS devices. The vulnerability, dubbed "Probllama" by the cloud security company Wiz, stems from insufficient input validation. Exploitation requires the threat actor to send specially crafted HTTP requests to the Ollama Application Programming Interface (API) server.

Submitted by Gregory Rigby on
Subscribe to