Apple extended patches for an actively exploited flaw to older devices, while CISA confirmed exploitation of two PHP2Shell vulnerabilities.
Cybersecurity6 items
Australia also reported disrupting a criminal's access to 180,000 Amadey-infected computers.
New5
Apple extends exploited CoreGraphics fix to older iPhones and iPads
Apple backported fixes for CVE-2026-86950, an actively exploited CoreGraphics vulnerability rated 8.8, to older iPhones and iPads. Processing a maliciously crafted file could trigger arbitrary code execution; Apple's October update release addressed 304 vulnerabilities overall.
Why it matters: Older devices now have patches available for a flaw already used in attacks, making update coverage an immediate priority.
Australia cuts Amadey operator's access to 180,000 infected computers
The Australian Signals Directorate said its operators accessed a criminal's Amadey backend and disconnected its controller from 180,000 infected computers worldwide. The disruption prevented that criminal from accessing or stealing data from those machines.
Why it matters: The operation interrupts access at substantial scale, though the report does not establish that the infected computers were cleaned.
Ongoing campaign uses external Teams chats for rapid malware delivery
A wide-scale, ongoing social engineering campaign combines mail bombing with external Teams chats to deliver malware in 15 minutes. The activity might be tied to the Black Basta ransomware gang.
Why it matters: The campaign makes external collaboration channels a relevant exposure for defenders investigating email floods and suspicious support approaches.
Aokigahara stealer targets developer keyrings and AI assistants
The Aokigahara information stealer targets developer keyrings and AI assistants. Its focus on AI tools can expose AI sessions, sensitive business data and developer infrastructure from a single compromised endpoint.
Why it matters: AI assistant sessions expand the sensitive material attackers can collect from compromised developer machines.
Google compares BigSleep and Fuzzilli vulnerability discovery in V8
Project Zero compared Google's LLM-based BigSleep system with the Fuzzilli JavaScript engine fuzzer. Over six months, BigSleep found 45 vulnerabilities versus Fuzzilli's 37, with a reported 25% overlap in the specific bugs found.
Why it matters: The limited overlap suggests AI-based discovery and conventional fuzzing can provide complementary coverage rather than replace one another.
Follow-ups1
CISA confirms active exploitation of two PHP2Shell vulnerabilities
CISA added two recently patched PHP2Shell flaws to its Known Exploited Vulnerabilities catalog, confirming active exploitation. Both vulnerabilities carry CVSS scores of 9.2 and enable unauthenticated remote code execution on affected deployments.
Why it matters: Official confirmation strengthens the case for prioritizing remediation of affected PHP deployments.