Cycle on Sep 21, 2026

I'll start by reviewing the open threads and existing notes for this topic, then research the latest developments in global AI liability regulation.

I have a clear picture of the topic's state: extensive coverage through mid-September 2026, with 11 open watch threads — several with September deadlines (SB 947 signing by Sept 30, OpenAI forum non conveniens motion filed Sept 2, Kistler MTD order pending). Today is Sept 21, so I'll check which watches have fired and scan for new developments this past week.

SB 947 is still on Newsom's desk (deadline Sept 30 — watch hasn't fired), and OpenAI's forum non conveniens motions are filed but no ruling yet. Let me check the other pending watches and sweep for new developments.

Two important leads: Judge Cannon granted Florida's motion to remand (Sept 8) — that fires a watch — and SB 947 still awaits action. Let me fetch the key articles to confirm details.

Confirmed: Cannon's remand ruling (Sept 8) fires the Florida watch, and the Tumbler Ridge FNC motions are fully briefed but unruled. Now let me check California's already-signed AI bills, the BC sovereign suit status, and EU developments.

California's picture is now clear — Newsom signed several chatbot/audit bills Sept 10, with SB 947 and others pending until Sept 30. BC's suit still not filed. Let me check EU developments and the appellate copyright ruling.

The 9th Circuit ruling on AI training copyright (Sept 16) is a significant developer-side win. Let me fetch the Reuters piece and check the EU omnibus adoption status.

EU omnibus was formally adopted (Parliament June 16, Council June 29, OJ July 24) — that updates prior findings. Now let me check the Meta layoff litigation watch and the new "pace AI" antitrust suit.

A genuinely new subject emerged: the Buist v. Anthropic "AI pacing" antitrust suit filed Sept 18. Let me verify the EU omnibus formal adoption, the Italy ad-ban decree, and the Kistler docket status.

Kistler has no ruling yet (watch stays open), and the EU omnibus timeline is confirmed: EP vote June 16, Council June 29, OJ publication July 24, in force July 27. Let me check the Italian ad-ban enforcement story before writing up findings.

All evidence is in hand. One more fetch to properly source the new antitrust suit, then I'll write the findings.

All sources verified. Now writing the findings — first, the two litigation updates (Florida remand; Tumbler Ridge forum fight).

Now the California legislative slate update and the 9th Circuit copyright ruling.

Now the two new subjects: Italy's landmark liability decree and the "AI pacing" antitrust suit, plus the EU omnibus formal-adoption update.

All findings written. Now the bookkeeping: record the fired Florida watch, register two new watches (Buist litigation, Italy decree entry into force), and open a thread on the pending GDPR/ePrivacy omnibus strand.

The cycle is complete. Here's what it surfaced: ## This cycle's findings (7 notes: 5 updates, 2 new subjects) **Watch fired — Florida AG v. OpenAI (resolved):** Judge Cannon granted Florida's motion to remand on Sept 8, sending the first state-AG sovereign AI enforcement suit back to Highlands County state court. She rejected both removal theories — the COPPA reference wasn't a "substantial federal question" under *Grable*, and COPPA doesn't completely preempt FDUTPA. The aggressive claims (strict liability, failure to warn, public nuisance) now head to a state-court jury. **Updates to standing findings:** 1. **Tumbler Ridge forum fight** — OpenAI filed identical forum non conveniens motions in a subset of the 37 federal suits, arguing B.C. is the proper forum ("California is not the most appropriate forum for claims by British Columbians attacked by a British Columbian in British Columbia"). No ruling yet; BC's sovereign suit still unfiled. 2. **California slate** — Newsom signed SB 1119 ("Adam's Law" child-safety audits), SB 867 (AI companion toy ban), and auditor-capacity bills on Sept 10; SB 947 (No Robo Bosses) and ~10 others await action by Sept 30 — including AB 2575, which would strip the "human override" defense from clinical-AI lawsuits. 3. **Ninth Circuit, Doe v. GitHub (Sept 16)** — DMCA claims against Copilot/Codex dismissed on appeal (outputs "never contained" CMI), but open-source license claims proceed; first appellate ruling in the AI-training wave. 4. **EU Digital Omnibus on AI is now law** — EP vote June 16, Council June 29, in force July 27; Annex III high-risk obligations deferred to Dec 2, 2027. **New subjects:** 5. **Italy's Decree 160 (published Sept 15)** — the first national framework converting AI Act obligations into criminal liability (1–5 years for omitting security/human-oversight measures, reaching *deployers*), corporate ad-ban sanctions, a presumption of causation, deemed-admission for missing logs, and direct action against insurers. 6. **Buist v. Anthropic (filed Sept 18)** — Sherman Act §1 class action alleging Anthropic, OpenAI, SpaceXAI, and Google illegally agreed to "pace the frontier" — making industry safety coordination itself litigable as an output-restraining cartel. **Bookkeeping:** resolved the Florida watch; registered new watches on the Buist litigation (ongoing) and Italy's decree entry into force (~Sept 30); opened a thread on the pending GDPR/ePrivacy omnibus strand. Unfired watches remain open: SB 947 decision, Tumbler Ridge FNC ruling, Kistler MTD order, BC filing, Connecticut (Oct 1), EU PLD transposition (Dec 9), and the 2027 effective dates.

The research cycle is complete. All findings are persisted under the topic, the fired watch is resolved, and the new forward-looking items are registered for future cycles to track.