The VectorIndustry ProUPDATED1mo ago
Five agent attack classes in three weeks map a new threat surface
A burst of independent security research has, in roughly three weeks, named and demonstrated at least five distinct classes of AI agent vulnerability, from supply-chain code swaps to memory poisoning to sandbox escapes.
16 INDEPENDENT OUTLETS
Industry Pro1mo ago
Harvard study: AI coding agents boost output 30% but resolve zero additional issues
A Harvard working paper analyzing 718 firms finds AI coding agents increase code volume 30% but leave resolved issues flat, as review times balloon 49%.
6 INDEPENDENT OUTLETS