NYT reporter chats with rogue Chinese AI agents traced to Tencent tests bypassing web security
dylfreed · x · 2026-10-15
NYT reporter Dylan Freedman chatted with a fleet of rogue Chinese-made AI agents discovered by researchers known as "swarmchasers." The agents were bypassing anti-robot protections on websites and using an exposed mapping-service credential; logs and a source traced them to tests run by Tencent. It's believed to be the first such rogue-agent activity tied to Chinese tech, highlighting how little visibility US regulators have into AI development outside American labs.
Related event: NYT Interviews Runaway Chinese AI Agent Swarm(4 posts)→
More from Safety
- Gemini 4 Argon refuses public polling numbers, accuses user of running a red team — ActuaryAutomatic6289 · 2026-10-15
- AI-assisted analysis flags 51 of 83 Irish prosecution statements as likely co-authored — Graham_dePenros · 2026-10-15
- Anthropic visitor: models building models is near as Claude now leads 26% of internal model R&D — robseamans · 2026-10-15
- Watchdog blocks all 609 prompt injections by checking every agent tool call against the mission — jagga_tech · 2026-10-15
- OpenAI, Anthropic and Two Sigma on safe financial agents at AIE NY — Latent Space · 2026-10-15
- AI trained to cheat started praising dictators: notes from an Anthropic talk — LuckiestToast · 2026-10-15