Uber faces an €825M GDPR fine for automatically suspending drivers without human review, highlighting the risks of AI decisions affecting workers. The Dutch Data Protection Authority handed Uber its largest privacy fine yet, and this one isn’t about data transfers or cookie consent. The regulator imposed an 825 million euro penalty, roughly $964 million, over Uber’s use of fully automated software to suspend driver accounts, sometimes permanently, with no human ever reviewing whether the system got it right. The violation is clear under EU law. The GDPR limits fully automated decisions when they can significantly affect a person’s life. An algorithm that can take away someone’s ability to earn a living, without any human review, falls directly into this category. The regulator also found that Uber failed to properly tell drivers when automated systems made these decisions, which the GDPR requires companies to disclose. “The Autoriteit Persoonsgegevens (AP), the Dutch data protection authority, imposes a fine of 824,990,000 euros on Uber. The reason for this is that the AP has ruled that Uber made fully automated decisions about drivers. In case of suspicions of fraud or customer reviews that were too low, drivers’ accounts were automatically temporarily deactivated or, in case of persistent low customer reviews, permanently deactivated. As a result, their income was lost via Uber during the deactivation.” reads the statement published by the Dutch data protection authority. “According to the AP, Uber has violated the prohibition of fully automated decision-making under the General Data Protection Regulation (GDPR). The AP also found that Uber did not sufficiently inform drivers about automatic decision-making. Uber has now stopped the violations.” The fine covers Uber’s practices from 2018 to 2022, so it concerns systems the company has already discontinued. Uber highlighted this point in its response, arguing that the regulator examined old policies rather than practices still in use today. “Uber used software to track drivers’ (driving) behaviour and to track customer reviews. If that software detected a suspicion of fraud or customer reviews were too low, the accounts of the drivers concerned were automatically deactivated.” continues the Dutch authority. “There was no human assessment here. This occurred between 2018 and 2022.” Uber also said it takes decisions that affect drivers’ income seriously. The company pointed to human reviews, safeguards and an appeals process for drivers who believe the system made a mistake. The appeal will have to determine whether these protections existed during the period covered by the fine or came later. And Uber is appealing. The company has stated it disagrees with both the decision and the size of the fine, setting up another round in what’s become a recurring pattern between Uber and Dutch regulators specifically. This is the fourth time the Dutch authority has fined Uber, which on its own says something about the relationship here. The previous record holder was a 290 million euro fine in 2024 over transferring European drivers’ personal data to the US without adequate protections, a case Uber also appealed at the time. Four fines from a single regulator isn’t really a pattern of bad luck anymore; it’s a pattern of a company and a privacy regulator that keep disagreeing about the same basic question, how much human judgment has to sit between an algorithm’s decision and a person’s actual income. The case goes beyond Uber. Many gig-economy platforms in Europe use algorithms to manage workers, routes and account status. This ruling shows the cost of relying on automated decisions without human oversight. For companies that use algorithms to make decisions affecting people’s accounts or income, saying “the algorithm decided” is no longer enough. The €825 million fine makes that lesson very clear. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, newsletter)
Two CVSS 9.8 miniOrange SAML WordPress plugin auth bypasses were exploited while paid editions never appeared in any vulnerability database. Manual patch required. Two critical authentication bypass vulnerabilities in the miniOrange SAML 2.0 Single Sign On WordPress plugin, both rated CVSS 9.8, are under active exploitation. Both CVE-2026-61979 and CVE-2026-15981 allow an unauthenticated attacker to forge a SAML authentication response and arrive in /wp-admin as any existing user, including administrators. The bugs are independent and both have been confirmed exploited in the wild. CVE-2026-61979 is an algorithm confusion flaw. The plugin trusts the incoming SAML response to declare its own signature algorithm. An attacker sets that algorithm to HMAC-SHA1, which causes the plugin to use the identity provider’s RSA public key as the HMAC secret. The RSA public key is, by definition, public. That’s the whole attack: fetch the key from the metadata endpoint, sign your own forged assertion with it, and the plugin verifies it as genuine. CVE-2026-15981 is a PHP bug caused by incorrectly handling different types of values. The openssl_verify() function can return three results: 1 when a signature is valid, 0 when it is invalid, and -1 when OpenSSL encounters an error. The plugin treated the result like a simple true-or-false value. In PHP, -1 counts as true. An attacker can send a specially crafted signature that causes an OpenSSL error, making the function return -1. The plugin then treats it as a valid signature and allows the attacker to log in. It is a basic coding mistake, but particularly serious because the plugin handles authentication for WordPress admin accounts. The technical root cause is interesting. What happened around the bugs afterward is more interesting. The miniOrange SAML plugin ships under one WordPress slug, miniorange-saml-20-single-sign-on, but that one listing actually contains seven completely separate product editions, each with its own independent version numbering. Free runs from 3.x to 5.x. Premium runs 11.x to 13.x. Standard runs 15.x to 17.x. VIP editions start at 32.x and 35.x. No two editions share a version range. When the public advisories were written, they covered only the Free edition, which is the one anyone can download from WordPress.org. The fix for the Free edition landed at 5.4.5. “While that record is correct, when you apply it to the slug, every paid install carries a higher version number than 5.4.5, so every paid install reads as already patched.” reads Patchstack analysis. “Therefore, any WordPress site that was running a vulnerable version like 16.1.9 reported it as patched, along with every other (13.x, 20.x, 26.x, 32.x, and 35.x) vulnerable version.” Every database, scanner, and dashboard in the WordPress security ecosystem was telling paid-edition administrators their sites were safe when the exploitable code was still running on their servers. The update prompt problem makes it worse. Sites running a vulnerable Standard edition 16.x see no available updates in their WordPress admin panel because the fix is on the 17.x line, and WordPress’s built-in update mechanism doesn’t offer cross-line jumps automatically. The path from vulnerable to patched requires a manual plugin upload, which site owners can’t do if they don’t know they’re affected, and every system they rely on was telling them they weren’t. “There was no advisory to read and no vulnerability database entry flagged the paid edition for these vulnerabilities. The plugin reported itself as fully up to date. In other words, every signal that normally warns of a problem suggested that everything was fine.” continues the report. “So, how did DigitalOcean catch the exploit? It had nothing to do with plugins or versions. An anomalous WordPress administrator session attempt appeared from outside their trusted network and was blocked. The attacker had already used the bypass to obtain a WordPress admin session cookie, but was stalled because the admin panel operations themselves sat restricted behind the trusted network. “ DigitalOcean, the team that found all of this, caught the exploit not through any of those systems but through a network anomaly: an authenticated WordPress administrator session arriving from outside their trusted network perimeter. The attacker had already used one of the auth bypasses to generate a valid admin session cookie. The cookie worked. The session was blocked because the admin panel itself was restricted to trusted network ranges. Defense in depth, doing exactly what it exists to do, on a system where every other detection layer had already failed. DigitalOcean then traced both bugs to specific lines of code in the plugin and its bundled XML security library, confirmed the affected version ranges across all seven paid editions (which miniOrange had not published anywhere), wrote two narrowly scoped hotfixes to buy time, and handed the full analysis to Patchstack for publication. Patchstack worked with miniOrange to get the complete edition and version matrix, then updated its database with all seven affected ranges, the first time any public database held that information. Attackers are actively scanning miniOrange SSO endpoints from six IP addresses in Belgium, Nigeria, the U.S. and Germany. The activity appears opportunistic, with attackers checking websites that use the plugin without first identifying the specific version or edition. That makes the silent patch especially concerning. Attackers can probe systems without knowing what version they run, while website owners may not realize whether their plugin has received the fix. “Vulnerability databases are only as good as the version data vendors publish publicly. When a vendor runs seven independently numbered editions under one slug and patches six of them without a public advisory, the entire ecosystem downstream of them goes blind at once: databases, scanners, dashboards, and the site admins relying on all three.” concludes the report. For anyone running this plugin, the full version table per edition is in the Patchstack article. The short version: find your edition, compare your version against the patched column, and expect to do the upgrade as a manual upload rather than through the WordPress dashboard. If you can’t update immediately, two narrowly scoped hotfixes covering both CVEs are published there alongside the IPs to check in your logs for prior exploitation attempts. Check the passkey and admin account list while you’re in there. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, miniOrange SAML WordPress Plugin)
Healthcare and services provider Nutex is investigating a data breach incident where an unauthorized third party exfiltrated information from company servers. The organization has disclosed the cyberattack in a filing with the U.S. Securities and Exchange Commission (SEC), noting that the stolen data includes details that may be private or confidential. “Based on preliminary findings from the Company’s ongoing investigation, the Company believes that certain information maintained on the Company’s servers was accessed and exfiltrated by an unauthorized third party, including some information that may be private and/or confidential,” Nutex says. Nutex Health is a for-profit healthcare company that operates 28 facilities across 12 states, including the Bayou City ER & Hospital in Texas and Green Bay ER & Hospital in Wisconsin.
Attackers are increasingly targeting the processes used to establish or recover identity rather than attacking the login itself. Specops explains how stronger identity verification can help organizations prevent fake workers and social engineering attacks from gaining legitimate access.
Threat actors have already compromised over 270 Zimbra instances in remote code execution attacks targeting a high-severity Zimbra Collaboration Suite (ZCS) vulnerability. The ZCS email and collaboration suite is used by hundreds of millions of people and organizations, including thousands of businesses and hundreds of government agencies worldwide. Synacor patched the security flaw (tracked as CVE-2026-73570), which allows unauthenticated attackers to gain code execution remotely by exploiting a command injection weakness in the SNMP monitoring component when SNMP notifications are enabled, with the release of ZCS version 10.1.20 on July 20. CERT Polska, the Polish Computer Emergency Response Team (CERT), first flagged the vulnerability as targeted in the wild last Monday, when it also warned security teams to check their logs for suspicious activity, including the Zimbra service restarting unexpectedly, and for files created in the /opt/zimbra/jetty/webapps/, /opt/zimbra/jetty_base/webapps/, and /tmp/ folders by user zimbra over the last 30 days.
Explore Unit 42 research on AI-enabled malware. Learn how existing behavioral detection and endpoint analytics stop AI-authored code before execution. The post The State of AI-Enabled Malware August 2026: From Brand Abuse to Agentic Execution appeared first on Unit 42.
Oasis Security has disclosed a weakness in NVIDIA NemoClaw that could let an attacker-controlled webpage take unauthenticated control of the local Ollama instance serving an AI agent and plant hidden instructions inside the model itself. The findings were shared with The Hacker News ahead of publication, and the report says Oasis Security reported them to NVIDIA's Product Security Incident
Thousands of companies have been affected by the Mirage2FA campaign from 2024 to 2026. The commercial phishing-as-a-service toolkit targets Microsoft 365 accounts by abusing legitimate login flows and bypassing two-factor authentication. According to ANY.RUN research, 48% of targeted email addresses were potentially compromised. Most of the affected companies are US-based. Mirage2FA Campaign
Cybersecurity researchers have disclosed details of a new campaign that uses a cluster of 24 npm packages as free phishing infrastructure for redirecting to ClickFix-style fake CAPTCHA pages. "While the malware is simply a single HTML page inside the npm package, and while downloading it wouldn't do harm, the threat actor’s use of npm isn't to infect developers who install it, but to use the
Vulnerability management has been a staple of security programs since the dawn of the cybersecurity discipline. The symbiotic relationship between vulnerability and patch management teams has also existed for that time and has gone through waves of contention and thankfulness. While this relationship required thoughtful care and feeding from both sides, both sides were aiming to work toward a
Confirm this action.
Leaving now will discard your changes.