AI Law Wiki News for October 15, 2026 — Episode 18
Ep. 18

AI Law Wiki News for October 15, 2026 — Episode 18

•

Episode description

AI Law Wiki News for October 15, 2026 — Episode 18

The FTC demands AI-safety records from OpenAI and Anthropic. Also: global and Hong Kong banking guidance, Apple’s trade-secret hearing, expert discovery, Denmark’s digital-likeness bill, and election safeguards in Britain and Brazil.

This report and the anchor voices were created with generative AI. Wadilya Saw is voiced by ElevenLabs Matilda and Max Token by ElevenLabs Rob, using Eleven v4. The established announcer and Craig’s licensed music, The Hour Commences, are retained.

October 15 news archive

1. FTC demands OpenAI and Anthropic safety records

Civil investigative demands delivered October14 seek evidence in a consumer-protection investigation. CBS confirms the subpoenas; Axios details their scope and reported testimony plans. OpenAI says it will engage constructively. No liability finding is reported.

Axios CBS News

2. Global financial watchdog strengthens oversight guidance for AI agents

The October15 final toolkit contains twelve practices for financial institutions. Reuters reports stronger human oversight for higher-risk agents and risk-based triggers for human approval. The practices are guidance adapted to institutional risks.

Financial Stability Board Reuters / Yahoo News

3. Hong Kong launches ART framework for responsible banking AI

The October15 ART framework accompanies updated supervisory guidance, a training-data sharing platform and sandbox testing. A future discussion paper will explore a material-agent registry and intervention mechanisms.

Hong Kong Monetary Authority Hong Kong banking AI guidance

4. OpenAI agrees to refrain from using Apple’s alleged trade secrets

9to5Mac, citing Bloomberg, reports an October14 voluntary undertaking concerning Apple’s alleged trade secrets and forensic images from fourteen employees’ devices. The judge indicated a preliminary injunction was likely; the report does not establish issuance.

9to5Mac

5. Nevada court distinguishes AI research from protected expert preparation

Hunton Andrews Kurth’s October15 account describes a September29 Nevada ruling distinguishing expert facts/data considered from protected strategy, deposition preparation and draft material. It requires AI case summaries to be disclosed for scrutiny, without finding that they were fabricated.

Hunton Andrews Kurth

6. Denmark introduces bill against nonconsensual digital imitations

The culture ministry’s October8 proposal would protect realistic nonconsensual digital imitations while preserving qualified exceptions for parody, satire and public criticism. Turn-of-year commencement is the minister’s aim, not an enacted date.

Danish Ministry of Culture

7. UK election regulator develops rapid deepfake alerts

Tech Policy Press reports a three-hour notification target and a May pilot: about three million posts,147 human-review flags and ten misleading-item referrals. The project focuses on election administration.

Tech Policy Press

8. Brazil investigates election-content restrictions on Meta and X

An October14 administrative inquiry asks Meta and X to explain reported election-content restrictions within five days. Separately, Lula’s campaign petitioned the electoral court over alleged unequal MetaAI image restrictions. Meta attributes temporary candidate restrictions to election rules.

Federal Public Prosecutor’s Office Agência Brasil

0:00Opening
0:45FTC seeks AI safety records
2:27Financial supervisors address AI agents
4:35Apple and OpenAI trade-secret hearing
5:45What an expert must disclose about ChatGPT
7:08Denmark proposes protection for digital likenesses
8:07Election alerts and platform restrictions
9:53Closing