Cybersecurity Threat Landscape
54 reports
Microsoft Patches “Reprompt” Flaw: Single-click Copilot Vulnerability
Researchers uncovered a vulnerability in Microsoft Copilot which allows threat actors to exfiltrate chat histories via a single malicious link. The flaw enables silent command execution within browser-based sessions to send sensitive data to external servers without further user interaction.
Malicious Chrome Extensions Impersonating AI Tools Steal User Data
Two malicious Chrome extensions, "Prompt Poaching," were found harvesting ChatGPT and DeepSeek chat histories. Posing as productivity tools, they bypassed security—one even holding a "Featured" badge—to exfiltrate sensitive AI data and intellectual property.
Threat Actors Exploit Google Email Cloud Features In ‘Multi-Staged’ Phishing Attacks
Threat actors are abusing Google Cloud’s “Application Integration” tool to send phishing emails from a genuine Google address. By bypassing filters and using fake CAPTCHAs, the multi-stage attack tricks users into entering Microsoft 365 credentials on fraudulent login pages.
Clickfix Attack: Fake BSOD and Booking Targeting European Hospitality Sector
The ClickFix campaign "PHALT#BLYX" targets the hospitality sector using fake BSOD errors to trick users into executing malicious code. By following "fix" instructions, victims paste commands into the Windows Run box, installing the DCRAT trojan to grant attackers remote system control.