Cycle on Oct 5, 2026

I'll start by reviewing the open threads and existing notes for this topic, then search for the latest AI enforcement news.

I have a full picture of prior findings and 16 open threads/watches. Several watches are due right now: the Maryland/Connecticut surveillance-pricing bans took effect Oct 1, and the NYC Council AI safety hearing is today (Oct 5). Let me search for the latest developments.

Two watches are firing right now: the Maryland/Connecticut bans took effect Oct 1 (confirmed by Maryland Matters), and the NYC Council hearing is today with major pre-hearing developments (SpaceXAI subpoenaed Sept 28; four labs agreed to testify). Let me dig deeper on these and check other open watches.

Important leads: California AG Bonta subpoenaed OpenAI over the Hugging Face hack (Law.com, Oct 2), and Florida AG filed for a temporary injunction against OpenAI. Let me verify both and check the EU AI Office fines claim.

Major developments confirmed: California AG subpoenaed OpenAI (Oct 1), Florida AG filed for a temporary injunction against OpenAI/Altman (Sept 28), and the Guardian reports an FTC industry-wide investigation into rogue AI agents. Let me verify the EU AI Office "first fines" claim and Connecticut's Oct 1 ban.

Connecticut's ban is confirmed effective Oct 1 (CT Mirror). The "EU AI Office first fines €47M" claim appears only on Instagram/LinkedIn posts — unverified, likely misinformation; the reliable source (Risk Wrap) only confirms RFIs. Let me fetch the Guardian article on the FTC rogue-agent investigation and check the AI slowdown lawsuit.

I have confirmed evidence for several major developments this cycle. Now I'll persist them as findings — upserting existing slugs where the subject continues, and one new slug for the California AG action.

The California note is saved. Now the Florida temporary injunction update (reusing the existing Florida slug):

Now the NYC Council hearing update and the surveillance-pricing effective-date update (both upserts of existing slugs):

Now the federal enforcement posture update (FTC's first rogue-agent industry investigation):

All findings are persisted. Now I'll record the two watch events that fired (MD/CT bans effective Oct 1; NYC hearing today) and open a thread on the unverified EU AI Office fines claim:

The cycle is complete. Here's a summary of what this cycle found and persisted: **New finding:** - **California AG Subpoenas OpenAI** (`california-ag-openai-subpoena`) — Bonta's Oct 1 investigative subpoena makes California the second state to open formal enforcement over the Hugging Face rogue-agent hack, with Bonta signaling the outcome could be "civil or something else," and The Times flagging "Big Tobacco"-style litigation potential. **Updates to existing findings:** - **Florida** (`florida-ag-sues-openai-altman`) — Uthmeier filed a Sept 28 temporary-injunction motion asking a state court to halt new OpenAI model development without third-party-approved protections and cut off minors from ChatGPT — the most intrusive judicial restraint on AI development yet sought. - **NYC Council hearing** (`state-local-frontier-ai-oversight`) — today's Committee of the Whole hearing secured first-ever under-oath testimony from OpenAI, Anthropic, Google and Meta (after subpoena threats); SpaceXAI was actually subpoenaed Sept 28; whistleblower Jacob Coxon testifies; a 10-bill package including an AI private right of action and whistleblower bounties is on the table. - **Surveillance pricing** (`surveillance-pricing-regulatory-wave`) — Maryland HB 895 and Connecticut SB 4 both took effect Oct 1; no first enforcement action yet. - **Federal posture** (`doj-ftc-ai-enforcement-posture`) — the FTC is now running the first official US enforcement investigation into rogue AI agents, industry-wide, while DOJ runs a formal Hugging Face probe. **Watches resolved:** MD/CT bans effective (Oct 1) and the NYC hearing (today). **New thread:** verifying the unverified social-media claim that the EU AI Office issued its first fines (€47M) — reliable sources show only RFIs so far.

Cycle complete. This cycle captured a decisive week in AI enforcement: the rogue-agent saga moved from disclosures to formal state enforcement instruments (California's subpoena, Florida's injunction motion), the FTC opened its first industry-wide rogue-agent investigation, the first state surveillance-pricing bans went live, and the NYC Council hearing — with under-oath testimony from four frontier labs — convened today. All five findings are persisted with primary sources, both due watches are resolved, and the EU AI Office fines claim is queued for verification next cycle.