grep -rl "ai" ./articles
The claim is everywhere and the argument behind it is good. The number is not a measurement. It comes from a single blog post in May 2026 describing a local experiment on React, and the industry statistics stacked around it come from different vendors counting different things.
The 'AI slop' story says quality collapsed. HackerOne's own figures say the confirmed-exploitable rate held at about 25% through a record surge, and the share of critical and high findings went up. The problem is not slop. It is that finding got cheap and fixing did not.
Sequoia, Khosla and Coatue backed a 2025-founded company whose pitch rests on one test: general-purpose models could run end-to-end attacks and failed to defend against the same attacks.
'AI Sidebar with DeepSeek, ChatGPT, Claude and more' had 300,000 installs and a 4.6-star rating before Google removed it in January. It returned in August, clean for eleven days, then shipped a 21-line addition.
Genians found the stack on infrastructure linked to the North Korean espionage unit. Running models offline means stolen documents never touch a provider, and no usage policy applies. That is the point.
Ballena Azul did not exist. The recruitment did. Three developers came through a referral chain, signed contracts, got VM access — and within minutes ran dxdiag and checked which country their connection appeared to come from.
The week of 18–24 July 2026 was read as a physical-AI story, and Atoms' $1.7 billion made that easy. Underneath it, Cathedral, Glow and Neo Security took $360m between them, which says more about where security capital is going than any sector total does.
Nearly 140,000 US tech jobs went this year, and AI keeps appearing in the explanation. But only four employers said anything specific, and the Financial Times found firms citing AI underperformed the Nasdaq by around 10% in the following 30 trading days.