Cyber ·
Threat intelligence report of October 19: breaches, AI gateway flaws and phishing
Summary of the Check Point Research report of October 19, 2026 on data breaches, service attacks, critical flaws in an AI model gateway and phishing campaigns.
The October 19, 2026 report, published from the Check Point Research feed, gathers data breaches, service attacks, exploited vulnerabilities, phishing campaigns and research on AI-related threats. Cases cited include CenterPoint Energy, which reported a breach affecting more than 1 million Texas residents with names, addresses, Social Security information and driver's license numbers; Burger King Japan, where unauthorized access between October 2 and 7 reached about 11.18 million app members, with no evidence of compromised passwords or payment cards; and ConnectWise Cloud Backup, where compromised application credentials gave access to at least 173 organizations.
In AI, researchers identified critical flaws CVE-2026-35029 and CVE-2026-59823 in a widely used AI model gateway, which the vendor has fixed. According to the report, they could let low-privilege users change settings, access confidential files and steal credentials. In controlled experiments, repeated peer influence raised one agent's rule-violation rate from 44% to 98%. India's CERT-In reported unauthorized access to organizations through autonomous AI agents that exploited exposed credentials and misconfigured cloud services.
The available text is a machine translation and is cut off before the Microsoft vulnerability section. To verify the details, consult the original report indicated by Check Point Research and the regulatory notices cited by the companies.
Original source: 19 de outubro – Relatório de inteligência de ameaças