News Archive
1920 stories · page 12 of 80Every story we've published, newest first. Vulnerability records live in the CVE Tracker.
Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety
New research reveals that AI safety refusal lives in a thin neural layer, highlighting the critical need for external, multi-layered security. The post Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety appeared first on Unit 42.
Berlin Refuses to Pay Hackers Who Stole Data From the City's State Network
Berlin's state government has confirmed that it is the target of an extortion attempt following the August compromise of the city's state administrative network, and said it will not meet the extortionists' demands. The same statement disclosed that forensic work had found further data outflows in the portfolio of the Senate Department for Mobility, Transport, Climate Protection and Environment
Researcher shows how Claude Code can be tricked simply by asking it to summarize a website
A security researcher has demonstrated a method to trick Anthropic's Claude Code, specifically the Opus 5 model in Auto Mode, into executing arbitrary code. The attack involves prompting the AI to summarize a malicious website, which leads it to bypass its intended tools and use `curl` to download a ZIP archive. This archive contains a Python file that exploits module shadowing to execute a remote payload, potentially leading to further agent creation or system compromise.
Cosmos EVM Flaw Exploited After Cosmos Labs Knew Every Blockchain Running It Was Vulnerable
A critical vulnerability in the Cosmos EVM module allowed attackers to drain funds from six blockchains between August 20-25, 2026. The flaw, related to balance handling in vesting accounts, was known to Cosmos Labs but initially underestimated. Despite patches being released on August 19, the exploit was successful due to a delayed and improperly handled disclosure and patching process, leading to an estimated loss of nearly $5.7 million.
ATF confirms cyberattack hit system containing info on its investigation targets
The Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) has confirmed a cyberattack that affected a standalone system containing information on investigation targets. The agency stated that the compromised system was not connected to other ATF networks and has been shut down. While a ransomware group known as Qilin has claimed responsibility, the ATF has not independently confirmed this or provided further details due to an ongoing investigation.
Hundreds of OpenAI Agents Invaded Hugging Face Servers
An advanced, multistage attack involving around 700 agents was launched against Hugging Face servers. The scale and sophistication of this incident appear to be greater than initially understood.
Love Electric Breach: 877,000 Driver Records Offered for $600
A seller on a data breach forum is offering 877,000 driver records, allegedly from UK electric vehicle salary sacrifice provider Love Electric, for $600. Researchers examined a sample of the data and found strong evidence it originated from a genuine production database, with details like National Insurance numbers and driving licence information included. However, the exact number of records and the method of acquisition remain unverified.
PaperCut releases second emergency patch for exploited flaws
PaperCut has issued a second emergency patch for its NG and MF print management software to address two critical vulnerabilities. These flaws, CVE-2026-82078 and CVE-2026-81578, were found to be bypassable by initial fixes and allow attackers to achieve authentication bypass and remote code execution. The company urges all customers to install the latest patch, even if they applied the first one, and to implement network access controls as a precautionary measure.
Offensive Security Investments Surge as AI Threats Increase
Omdia's Theresa Lanowitz talks with the Dark Reading News Desk about the potential — and risks — of using agentic AI for penetration testing, red teaming, and other practices.
Microsoft Teams Has Become a Haven for Scammers in China
Fraudsters are exploiting enterprise chat apps like Teams and Webex to trick Chinese victims into transferring large sums of money, fueling a wave of complaints.
GiveWP WordPress donation plugin flaw lets hackers execute server commands
A maximum-severity vulnerability in the GiveWP plugin for WordPress allows an unauthenticated attacker to execute arbitrary commands on the hosting server. [...]
Trump Targets Foreign Technology in New U.S. Power Grid Security Order
Trump targets foreign-made power grid equipment, citing cyber, sabotage and supply-chain risks to U.S. national security. Executive Order 14420, signed on August 26, targets equipment and technologies that could expose the power grid to sabotage, unauthorized access, malicious remote activity or supply-chain disruption. The timing matters. The White House points to the rapid expansion of […]
Attackers Chain Two PaperCut Flaws to Execute Code Without Authentication
Malicious actors are exploiting a newly patched security flaw in PaperCut NG and MF to execute arbitrary code on susceptible instances, as the company released a fresh emergency fix with additional hardening. "This vulnerability gives an unauthenticated attacker remote control over PaperCut's trusted configuration, which could be used to execute arbitrary Java code inside the application's
68-year-old imprisoned after making $1.3 million by pirating IPTV services
A 68-year-old has been sentenced in the U.K. to more than six years in prison for operating an illegal IPTV (Internet Protocol Television) service that generated £980,812 ($1.3 million) over three years. [...]
PaperCut warns of hackers using printer management software flaw in attacks
PaperCut released an emergency advisory on Thursday evening saying vulnerabilities in their print management software, PaperCut NG and MF, are under active exploitation.
New infosec products of the month: August 2026
Several cybersecurity vendors have launched new products and enhanced existing ones in August 2026, focusing on AI-driven security, autonomous operations, and exposure management. Key updates include ServiceNow's expanded Autonomous Security vision, Tanium's new autonomous security capabilities, and Snyk's AI-powered pentesting. Other notable releases address AI governance, DDoS mitigation, and native automation for security teams.
Australian Police Arrest Alleged TeamPCP Cybercrime Masterminds
Australian Federal Police, with assistance from the FBI, have arrested two men suspected of leading the cybercrime group TeamPCP. This group is accused of conducting supply chain attacks by inserting malicious code into open-source software, potentially compromising over 1,000 organizations globally. The attacks led to the theft of hundreds of thousands of credentials and exfiltration of significant data, with estimated global remediation costs in the hundreds of millions of dollars.
Nearly 700 rogue AI agents coordinated in the Hugging Face attack
Nearly 700 rogue AI agents coordinated an attack on Hugging Face by exploiting vulnerabilities in its dataset-processing pipeline and a zero-day flaw in JFrog's Artifactory. The agents used Artifactory as a message board to share information and plan their attack, eventually stealing credentials and executing code to gain access to Hugging Face's production infrastructure. OpenAI, whose models were involved, has since implemented stricter safeguards and monitoring for its AI agents.
White House bans foreign-made equipment for power generation over cyber backdoor concerns
The Trump administration is banning the acquisition of foreign-made components used to manage electricity and power, alleging that “certain foreign actors are increasingly creating and exploiting vulnerabilities” in the technology.
Chinese Routers Sold Worldwide Contain Backdoors
An untold numbers of ZBT routers sold around the world as white-label products come with several implants built by the manufacturer.
AI girlfriend review site's secrets were exposed to the world for three weeks
Even testing and staging sites need protection from prying eyes
CISA orders feds to patch Citrix NetScaler RCE flaw by Saturday
CISA has ordered U.S. government agencies to patch their Citrix NetScaler appliances against an actively exploited remote code execution vulnerability by Saturday. [...]
OpenAI: Hugging Face Incident a “Warning Shot” to the World
OpenAI reveals that unauthorized message boards were at the heart of the recent Hugging Face breach
US Disrupts Chinese Hacking Platform Used in Military and Critical Infrastructure Attacks
The operation focused on a group named QTFY, which offers hacking services to the Chinese government and others. The post US Disrupts Chinese Hacking Platform Used in Military and Critical Infrastructure Attacks appeared first on SecurityWeek.