Where we are now (2026) — and the five patterns that predict what happens next

Lesson 5 of 5 in From Asimov to the AI Act: A History of AI Governance.

The 2026 snapshot, as of this module’s writing. In the US: Texas’s TRAIGA and California’s SB 53 took effect on 1 January 2026; Colorado, after delaying its 2024 act, repealed and replaced it with the disclosure-based SB 26-189 (May 2026) — a live lesson in how quickly “first comprehensive state law” can become “former law”; the White House sent Congress a legislative framework for federal AI law (March 2026) and issued EO 14409 on frontier-model cybersecurity (June 2026), while New York’s RAISE Act approached effect (~January 2027). In Asia: Korea’s AI framework legislation took effect in January 2026 — the first comprehensive national AI law in force in Asia. At the UN: the first Global Dialogue on AI Governance convened in Geneva (July 2026). The summit series reached New Delhi (February 2026) as the AI Impact Summit, with Geneva planned for 2027.

And in the EU: the AI Act reached general application on 2 August 2026 — but with an asterisk the size of a directive. The Digital Omnibus package deferred the high-risk obligations: Annex III systems to 2 December 2027 and Annex I product-embedded systems to 2 August 2028, with legacy-GPAI compliance due August 2027 and new NCII/CSAM-related prohibitions plus synthetic-content transparency duties landing 2 December 2026. The world’s most ambitious AI law is in force, partly applicable, and partly postponed — all at once. If you can explain that sentence to a client, you understand the state of the art.

The summit road: Bletchley to Geneva

  • 2025-02-01 — Frontier safety frameworks become table stakes:

    Following the Seoul commitments, major labs publish or update frontier safety policies (capability thresholds, evaluation gates, deployment mitigations) ahead of the Paris summit.

  • 2023-11-01 — Bletchley Park AI Safety Summit:

    28 countries + the EU — including the US and China — sign the Bletchley Declaration on frontier-AI risk. The summit series begins.

  • 2024-05-21 — Seoul AI Summit:

    Frontier labs sign safety commitments — publish risk frameworks or explain why not. The summit series turns from declarations to developer promises.

  • 2025-02-10 — Paris AI Action Summit:

    The series pivots from safety to action and investment; the US and UK decline to sign the final declaration — the divergence made visible.

  • 2026-02-19 — AI Impact Summit, New Delhi:

    The summit series lands in the Global South, centering development and inclusion; Geneva planned as the next stop (2027).

  • 2026-07-06 — First UN Global Dialogue on AI Governance:

    The Global Digital Compact’s forum convenes in Geneva — every state at one AI governance table for the first time.

Summit diplomacy at a glance — watch the nouns change
SummitWhenSignature outputThe framing

Bletchley Park (UK)

Nov 2023

Bletchley Declaration — 28 countries + EU, incl. US and China, acknowledge frontier risk

Safety — existential register, invitation-only urgency

Seoul (KR)

May 2024

Frontier AI Safety Commitments — 16 companies pledge published safety frameworks; safety-institute network launched

Safety + innovation — companies join states at the table

Paris (FR)

Feb 2025

Statement on inclusive, sustainable AI — US and UK declined to sign; €-scale investment announcements

Action — growth and sovereignty edge out risk

New Delhi (IN)

Feb 2026

AI Impact Summit — development, inclusion, Global-South priorities (outcomes: check current records)

Impact — benefits framing completes the pivot

Geneva (CH)

Planned 2027

To be seen — alongside the UN Global Dialogue machinery in the same city

Open question: does “safety” return to the title?

Eighty-four years from Asimov to the Digital Omnibus. Strip the dates away and five patterns carry all the predictive power this history offers:

  1. The pacing problem — law moves in years, capability in months; every framework is a bet on how to regulate a moving target.
  2. Incident-driven regulation — COMPAS → fairness rules; Cambridge Analytica → GDPR enforcement politics; ChatGPT → the GPAI chapter. The next statute is being drafted by the next scandal.
  3. The pipeline: principles → policy → practice — soft law is not failed hard law; it is hard law’s larval stage. OECD words of 2019 became EU obligations of 2024 and audit checklists of 2026.
  4. Brussels effect vs. deregulatory competition — first-mover rules pull global compliance toward them, unless a rival bloc makes absence-of-rules its competitive offer. 2025–2026 is history’s first live A/B test between the two.
  5. The pendulum — safety framings and innovation framings alternate with elections, incidents, and market cycles (watch the summit titles: Safety → Action → Impact). Professionals build programs that survive the swing: controls anchored in standards and contracts outlive controls anchored in any one administration’s executive orders.

Interactive sorting exercise: File each event in its era. If you can do this cold, you own the narrative arc of the field.

Tool: Regulatory Time Machine — Ride the Timeline Time-Machine: stop at six pivotal moments, make the policy call yourself, then see what history actually did.

Interactive checkpoint quiz (1 questions) — open this page in a browser to take it.