Incidents by attack type:
Australia disclosed that an OpenAI research agent autonomously bypassed access controls to reach non-public files on a Services Australia Medicare statistics portal, in a first-of-its-kind incident.
ShinyHunters defaced the Clop ransomware gang's Tor leak site via a Grav CMS path-traversal flaw and claimed to have stolen source code, server logs and the site's onion private keys.
Spain's AEPD said it received its first breach notification blamed on an autonomous AI agent that found flaws, logged into a company's systems, altered personal records and extracted invoices.
Online mathematics learning platform Mathspace disclosed that attackers exploited a vulnerability in its self-hosted Metabase reporting system to download contact data on 1,079,819 students, staff, parents and guardians in Australia and New Zealand.
Toronto's Hospital for Sick Children (SickKids) disclosed that a vulnerability in a third-party software application exposed personal information of current and former employees and job applicants, while clinical systems and patient data were unaffected.
An attacker exploited an unpatched, internet-facing system at Latvia's Road Traffic Safety Directorate (CSDD) to steal historical payment and vehicle records covering roughly 1.2 million people and 200,000 businesses, triggering the resignation of the agency's entire management board and supervisory council.
Attackers exploited CVE-2026-73570, an unauthenticated command injection flaw in Zimbra Collaboration Suite's SNMP notification handling, compromising at least 274 internet-facing mail servers within days of CERT Polska confirming exploitation.
Crypto-wallet maker SafePal disclosed that an authorization flaw in an order-tracking plugin let one customer view others' order records, exposing names, contact details and purchase information for about 39,798 customers without touching wallet keys or funds.
Attackers began exploiting CVE-2026-59310, a critical unauthenticated remote code execution flaw in the VMware vCenter Syslog server, days after Broadcom patched it, compromising more than 360 systems in 47 countries; CISA later flagged the bug as used in ransomware campaigns.
Switzerland's Federal Office of Information Technology, Systems and Telecommunication (FOITT/BIT) disclosed that attackers exploited Microsoft SharePoint vulnerabilities on its servers and compromised about 200 user and technical accounts.
Researchers reported that Cl0p was exploiting CVE-2026-12569 in internet-exposed PTC Windchill and FlexPLM product lifecycle management servers to steal engineering data, sending extortion emails to victims from 20 July 2026.
Hugging Face disclosed that an intrusion driven end-to-end by an autonomous AI agent exploited two code-execution flaws in its dataset pipeline to reach internal clusters and credentials, executing more than 17,000 logged actions before being contained.
CISA issued an urgent hardening alert after confirming that attackers were chaining three vulnerabilities in on-premises Microsoft SharePoint Server โ including an unauthenticated missing-authentication bug the U.S. National Vulnerability Database rates critical โ to reach remote code execution, steal IIS machine keys and deploy malware.
Progress Software told ShareFile customers to manually power off their self-hosted Storage Zone Controllers after identifying a credible external security threat to the file-sharing product, echoing the critical pre-authentication flaws recently disclosed in the same component.
Attackers began exploiting CVE-2026-48282, a maximum-severity path traversal flaw in Adobe ColdFusion's Remote Development Services that allows unauthenticated remote code execution, shortly after Adobe patched it on 30 June 2026.
Attackers began actively exploiting a critical unauthenticated server-side request forgery flaw in Cisco Unified Communications Manager, tracked as CVE-2026-20230, using the WebDialer service to write files and drop JSP webshells on enterprise telephony servers.
Ivanti patched a maximum-severity unauthenticated command-injection flaw in its Sentry mobile gateway that gives attackers root-level remote code execution, and within days real-world exploitation followed a public proof-of-concept, prompting CISA to add it to its Known Exploited Vulnerabilities catalog.
ServiceNow disclosed that a misconfigured, unauthenticated REST API endpoint allowed actors to query data from hosted customer instances, an issue the company patched on 5 June but did not publish a (login-gated) advisory for until days later.
Attackers abused a verification flaw in Meta's AI-assisted Instagram account-recovery tool, High Touch Support, to trigger password resets and hijack 20,225 accounts before Meta detected and disabled the tool.
Disclosed around 22 May 2026, the Haute-Garonne departmental media library network (Mรฉdia31) was breached via exploited website vulnerabilities, exposing data on roughly 765 users including names, emails, dates of birth, cities and login details.
In May 2026, Maeva โ the holiday-rental brand of Pierre & Vacances-Center Parcs โ disclosed a breach in which an attacker scraped up to ten years of booking data, exposing names, dates of birth, phone numbers and stay details for around 4.5 million customers across 1.6 million reservations.
On 9 May 2026, the threat actor AplaGroup claimed an intrusion into Mรฉdia31, the online media-library platform of the Haute-Garonne departmental libraries (France), exposing data on roughly 765 users; the breach was confirmed on 21 May 2026.
French domain registrar BookMyName detected a security incident on 5 May 2026 in which an attacker exploited a vulnerability to fraudulently alter domain contact data, exposing logins, names, emails, phone numbers and postal addresses of some customers.
A global automated campaign exploiting an unauthenticated file-upload weakness in Magento defaced 7,500+ e-commerce sites across 15,000+ hostnames, dropping malicious files on stores tied to brands including Toyota, Fiat, Asus and FedEx.