15

public signals

4

signal clusters

14

source links reviewed

Signal 1 · Strong signal

NYC and Los Angeles paused student AI. That is a freeze, not a design.

On Sept. 2 New York City barred student-facing generative AI for 2-K through eighth grade for this school year, banned companion chatbots for every student, and capped recommended screen time. High schoolers get five supervised pilots. Los Angeles Unified barred student generative AI on district devices. Teachers can still use it. A freeze can protect productive struggle. It can also teach kids that AI is a home privilege.

Selected sources

NYC Mayor's Office announcement
Education Week on NYC and Los Angeles
LAUSD moratorium, Education Week / Los Angeles Times
GovTech on the NYC pause
Plus 2 more in the full report.

Signal 2 · AI governance

GPUs think. CPUs act. The governance problem lives in the acting.

Wccftech described GPT-6 Astra as a cloud orchestrator that hires local computer-use agents. They click, type, and drive apps, then return finished work. That is architecture reporting, not an OpenAI primary. Keep the split. Discount the certainty. Write the four controls before any computer-use pilot: which systems, what it must ask, what record it leaves, and which machine it may sit on.

Selected sources

OpenAI's GPT-6 Astra spawns armies of agents that run on local CPUs

Signal 3 · Cybersecurity and privacy

PaperCut's third emergency patch is the news. The KEV clock is not a rumor.

PaperCut shipped Emergency Patch Release 3 on Sept. 1 after Release 2 left bypasses. CISA added CVE-2026-81578 and CVE-2026-82078 to the known-exploited catalog. Huntress said the chain needs no password, only a reachable host. If NG or MF is internet-reachable, restrict it and apply Release 3. A missing server.log is evidence, not comfort.

Selected sources

URGENT Security Advisory: PaperCut NG/MF (27 Aug 2026), updated through Release 3
Behind the scenes of the August PaperCut security incident
PaperCut NG/MF critical zero-day exploited in the wild
PaperCut issues emergency patches as threat actors target chained vulnerabilities
Plus 1 more in the full report.

Signal 4 · Agentic AI checkpoint

Agentic AI checkpoint: Hermes shipped a society of bots; OpenClaw went stable

Hermes Agent published stable v0.21.0 on Aug. 31: Bot Mode, peer DMs, cron memory, live-steered subagents, and a desktop browser the agent can drive. OpenClaw published stable 2026.9.2 on Sept. 5. Last week those OpenClaw tags were still betas. A stable tag is not unsupervised authority.

Selected sources

Hermes Agent v0.21.0 stable release (v2026.8.31)
OpenClaw 2026.9.2 stable release

Read the full Sunday Signal.

Question to carry: after Florida's Sept. 16 vote, which public AI or device decision will show a named owner, a student-facing boundary, a dated review, and a stop condition?