AI Security Risks Spread Across the Week

AI systems emerged as both attack surfaces and tools in cybercrime, from agent-triggered package theft to an AI support bug that enabled account takeovers. Those concerns recurred in stories about AI spying, AI-enabled scams, and US security disputes over Anthropic tools.

The week also produced a concrete exploit pattern as a PeopleSoft zero-day moved from disclosure to active attacks and data theft. Microsoft's nearly 200-fix release showed the broader pressure to patch known attack paths quickly.