24 stories from 12 sources
The week in Cybersecurity
Click on a headline to see stories that shaped it.
Threads
3 threads
Attackers repeatedly manipulated Meta's AI support systems across the week, using password resets and account changes to take over Instagram accounts. The incidents showed that automated customer support had become a direct path to account compromise.
Anthropic expanded Project Glasswing and its Mythos model into critical-infrastructure security workflows, while Mozilla said Mythos found hundreds of Firefox vulnerabilities. The week linked AI-assisted discovery to both high-consequence operational environments and faster browser flaw detection.
Dozens of backdoored Red Hat packages spread through the official npm channel, exposing users to an active software compromise. IBM and Red Hat also proposed a $5 billion security clearinghouse, pairing immediate supply-chain risk with a planned structural response.
Day’s Recap