TechNadu

The roundup shows cybercriminals racing to use AI while controlled models slip past systems built to contain them. AI Agents exploited vulnerabilities, reused credentials, created identities, and contacted people. These actions were neither planned nor approved, and evaluators did not detect every step immediately. Some reports identified missing live monitoring as a contributing factor. U.S. authorities delivered two decisive cybercrime enforcement outcomes.

PIA Q2 2026 report details 19 legal requests received Private Internet Access (PIA) received 19 legal requests between April and June 2026, but produced no VPN activity or connection logs in response, according to the company’s latest transparency report. The VPN provider published its Q2 2026 report on August 6, detailing the legal requests it received during the quarter as well as activity in its bug bounty and vulnerability disclosure programs.

Add as preferred source on Google Question: The Arctic Wolf 2026 Trends Report found that security teams spend roughly 13 to 15 hours per week on each major security function. Based on what you are seeing, which tasks can automation and AI redistribute, and what proportion of the workload could they realistically absorb? How close are security teams to achieving a meaningful reduction in workload?

UNC6671 remained active after the apparent retirement of the BlackFile extortion brand, instead expanding its operations under several new names such as Redact, Pink, Helix, and Falcon, indicating that the same threat cluster is likely behind all five brands.

A macOS malware campaign steals credentials and drains cryptocurrency wallets, tracing the infection to a ClickFix social engineering scam. A Huntress report details an incident uncovered during a June 2026 retrospective threat hunt on a system that had actually been infected three months earlier.

An ongoing email-driven adversary-in-the-middle (AiTM) phishing campaign compromises Microsoft 365 accounts to collect finance-related email, with hundreds of organizations targeted in July 2026 alone, with confirmed intrusions across a broad range of environments. Arctic Wolf Labs has disclosed the operation uses voicemail-themed phishing emails carrying Microsoft branding and fabricated caller ID, date, duration, and reference number details designed to manufacture urgency.

Add as preferred source on Google Jeremiah Fowler, Security Researcher and EU Training Consultant at Black Hills Information Security, details the risks surrounding an exposed Brazilian health database. He explains how people and organizations can respond to it and similar incidents. People can change passwords, but they cannot replace their fingerprints or faces. What can someone realistically do after biometric information is exposed? Unlike a password, biometric data is permanent.

Hackers launched a coordinated wave of sophisticated cyberattacks against major Wall Street financial firms and money managers in recent days, using vishing to impersonate employees and executives, according to people familiar with the matter. Some of the world's largest hedge funds, along with several private equity firms, were among the targets. Point72 Asset Management told investors on Wednesday that it had suffered a cyberattack, one of the sources said, cited by Reuters.

Meta disclosed Wednesday that one of its AI models hacked a third-party company during cybersecurity testing. The company said it was investigating an incident in which a misconfiguration by Irregular, an independent company that conducts cybersecurity evaluations for Meta, inadvertently gave one of its models internet access during testing.

Image Courtesy of Rawpixel Inter-Con Security Systems, one of the largest private security companies in the world, has had its data published after a "pay or leak" extortion campaign carried out by the ShinyHunters cybercrime group. Breach notification service Have I Been Pwned (HIBP) added the breach to its database on August 5, 2026, confirming 276,114 unique email addresses among the exposed records, which translates to a fraction of what the attackers originally claimed to have stolen.