Stack layer / Threat pattern
A Show HN tracker puts numbers on model staleness: Gemini 3.1 Pro shipped 13 months after its training cutoff
Show HN
Policy dependency / Stack layer
Converting GUI Trajectories Into Replayable MCP-Style Calls Instead of Unstructured Memories
arXiv 2609.16635
Policy dependency / Stack layer
Ordewell uses a read-only coding agent as planner, then spawns per-task agent sessions gated by a marker-based verdict engine
GitHub
Stack layer / Threat pattern
Emergence World ran 10 agents per world for 16 days and found no frontier model contained an injected attack — one acted on poisoned memory 46 hours later
arXiv
Stack layer / Threat pattern
The Internet Archive is now rate-limiting the Wayback Machine against high-volume automated traffic
Internet Archive Blog (572 points on Hacker News)
Stack layer / Threat pattern
TypeSafe AI ships Jev, a model that returns typed probabilistic values instead of text, at $0.042 per million input tokens and free output
TypeSafe AI
Policy dependency / Stack layer
Google DeepMind Launches the DeepMind Institute, an Essay Platform on AGI Run by Legg, Hassabis and Manyika
Google DeepMind (corroborated by Axios and TNW) / Hacker News (174pts)
Stack layer / Threat pattern
37,623 provenance-labeled agent PRs: Codex code was reverted half as often as human code, Devin's 31% more, and Claude Code PRs waited 12.6 hours for first review
arXiv 2609.17598