IITP hack exposes South Korea’s R&D security lapses, triggers data leak probe - CHOSUNBIZ
The Institute of Information & communications Technology Planning & Evaluation (IITP) is currently investigating a data breach caused by a security lapse in its personal information access log management system. This incident marks the second hacking of the organization in seven months, raising serious concerns regarding the agency’s cybersecurity oversight.
Suno Got Hacked is an article detailing how a security breach of the AI music generator revealed that the company trained its models by scraping content from platforms like YouTube, Deezer, and Genius. This incident occurs alongside ongoing copyright lawsuits from major music labels and criticism from artists regarding unauthorized AI training practices.
Patient records stolen in cyber-attack on Australian healthcare provider | Health | The Guardian
A significant cyber-attack on the healthcare provider Partnered Health has resulted in the theft of personal and sensitive medical records across 21 clinics. The organization has subsequently sought an injunction to prevent the misuse or publication of the stolen patient data.
The Kudankulam nuclear power plant in India experienced a data breach involving stolen files leaked by the World Leaks ransomware group. While officials claim the incident does not affect nuclear security, the breach involved blueprints and supplier data from a plant contractor.
Romania’s land registry hit by cyber attack, data allegedly for sale - Help Net Security
Romania’s National Agency for Cadastre and Land Registration (ANCPI) confirmed a cyber attack disrupted its e-Terra application, halting real estate transactions. While the agency claims data remains secure, a threat actor known as ByteToBreach is allegedly selling stolen citizen data and source code on the dark web.
1 in 3 AI Agents Have Security Flaws: Is InfoSec Ready for the Next Supply Chain Attack?Webinar.
Recent data reveals that 36% of AI agents contain security flaws, exposing organizations to significant software supply chain risks like credential theft and malicious prompt injection. This webinar explores practical strategies to help InfoSec professionals identify and mitigate these vulnerabilities before they are exploited.
CISA orders feds to patch actively exploited Oracle flaw by Saturday
CISA has ordered federal agencies to patch a critical Oracle E-Business Suite vulnerability, CVE-2026-46817, by Saturday due to active exploitation. This security flaw allows unauthenticated attackers to gain system control via HTTP network access.
Specter Turns Your Flipper Zero Into a Pocket Skimmer Detector
Specter is a new Flipper Zero application that transforms the device into a passive, listen-only tool for detecting active 13.56 MHz NFC readers. By monitoring for radio-frequency fields, it alerts users to potential skimmers or suspicious hardware through visual and haptic feedback.
OpenAI Unveils GPT-Red AI Model That Automatically Finds Prompt Injection Vulnerabilities
OpenAI introduced GPT-Red, an automated red-teaming model designed to identify and exploit prompt injection vulnerabilities through self-play reinforcement learning. By generating adversarial prompts, this tool enhances the robustness of production models like GPT-5.6 Sol against unauthorized data access and system manipulation.
WhatsApp GhostPairing Lets Scammers Hijack Accounts Without Stealing Passwords
GhostPairing is a social-engineering tactic where scammers hijack WhatsApp accounts by manipulating victims into approving a fraudulent linked device. To protect against this account takeover, users should regularly monitor their linked devices and remain cautious of requests to scan unexpected QR codes.
20+ Hijacked Government Websites Became an Attack Channel
More than 20 Brazilian government websites have been hijacked by the PhantomEnigma campaign to serve as malware delivery channels through trusted links and authenticated emails. This operation uses a modular backdoor to infect systems, putting public agencies and financial institutions at significant risk of data theft and unauthorized access.
A new AI penetration testing framework shifts focus from traditional infrastructure compromise to identifying how prompt injection and other adversarial inputs can lead to operational-objective violations. This approach assesses whether attackers can successfully steer AI systems to act against their intended mission, such as manipulating workflows or bypassing security protocols.
Zoom security advisory (AV26-707) - Canadian Centre for Cyber Security
The Zoom security advisory (AV26-707) outlines critical vulnerabilities affecting various Zoom Windows products, including the Meeting SDK and Workplace clients. Users and administrators should promptly apply the latest software updates to mitigate risks like improper input validation and privilege management issues.
New TELEPUZ Malware Spreads via ClickFix to Steal Data and Run Commands
The TELEPUZ malware utilizes ClickFix social engineering lures to deceive users into executing malicious code that steals sensitive data and runs unauthorized commands. This modular malware-as-a-service threat employs advanced obfuscation and defense evasion techniques to maintain persistence and establish communication with its command-and-control servers.
Sunsetting the Public AttackerKB Platform | Rapid7
Rapid7 is sunsetting the public AttackerKB platform on August 18 to centralize vulnerability intelligence and exploit analysis within its existing ecosystem. Users can continue accessing high-fidelity technical write-ups and data through the Rapid7 blog and the Vulnerability and Exploit Database.
Splunk security advisory (AV26-708) - Canadian Centre for Cyber Security
The Splunk security advisory (AV26-708) outlines critical vulnerabilities in Splunk Enterprise and Splunk Cloud Platform. Users and administrators are urged to review the provided links and apply the necessary security updates to address these risks.
Millions of Shark Robot Vacuums Vulnerable to Unpatched Remote Code Execution Flaw
Millions of Shark robot vacuums are vulnerable to a critical remote code execution flaw stemming from insecure AWS IoT Core configurations that allow attackers to access cameras, home maps, and Wi-Fi credentials. Because no official firmware patch is currently available, experts recommend isolating these devices on a separate network to mitigate potential unauthorized control.
n8n Token Exchange Flaw Could Let Attackers Log In as Users From Another Issuer
A token exchange flaw in n8n (CVE-2026-59208) allowed attackers to log in as other users by incorrectly matching JWT claims based solely on the user ID while ignoring the token issuer. Users of the workflow automation platform are encouraged to patch to version 2.27.4 or later, or disable the vulnerable Enterprise feature to mitigate the risk of unauthorized access.
Sandworm hackers have a CAPTCHA trick for Ukrainians | The Record from Recorded Future News
Sandworm hackers are targeting Ukrainians with ClickFix, a social engineering technique that uses fake CAPTCHA prompts to trick victims into executing malicious PowerShell commands. This method leads to the installation of malware that grants attackers persistent access to compromised systems and facilitates further cyber espionage.
The backlash against Flock cameras is spreading | Malwarebytes
The backlash against Flock Safety’s ALPR systems is intensifying as agencies increasingly respond to concerns regarding system inaccuracies, misuse by law enforcement, and unauthorized data sharing. These risks to privacy and civil liberties have led some jurisdictions to disable surveillance networks or terminate contracts with the vendor.
Hackers Actively Exploiting SonicWall SMA1000 0-Day Vulnerability in the Wild
Threat actors are actively exploiting SonicWall SMA1000 appliances through a critical SSRF vulnerability (CVE-2026-15409) and a local privilege escalation flaw (CVE-2026-15410) to gain root access and compromise corporate networks. Organizations must urgently patch their systems to the latest firmware versions to mitigate this zero-day threat.
EU orders Google to open Android and Search to AI rivals, including OpenAI
Under the Digital Markets Act, the European Commission has ordered Google to open its Android system and Search infrastructure to AI rivals like OpenAI and Anthropic by mid-2027. This mandate requires Alphabet to provide competitors with system-level access and search data to foster innovation and competition.
SEC proposes Regulation E-Delivery to modernize how investors receive securities information
The SEC has proposed Regulation E-Delivery to make digital communication the default for securities disclosures, replacing the current paper-first requirement. This shift aims to modernize investor communication and could save the industry approximately $800 million annually in printing and postage costs.
Next.js Launches Monthly Security Release Program as First Update Patches 9 Vulnerabilities
Next.js has introduced a monthly security release program to provide predictable updates, beginning with a patch addressing nine vulnerabilities across versions 16.2 and 15.5 on July 20, 2026. This initiative aims to improve vulnerability management by offering developers and partners advance notice to plan upgrades and implement necessary mitigations.
Moonshot AI has launched Kimi K3, a new model featuring a 1 million context token capacity and multi-agent swarm technology. This advancement intensifies the AI arms race and could significantly impact crypto markets by driving demand for decentralized compute and validating agent-based projects.