AI Ethics News — 2026-10-24 (Sat)

Overview

Questions about who has a say in AI at work (ILO) and whether chatbot conversations and children's data are protected (Mezha.net, Top Class Actions) surfaced over the past 48 hours. Fake AI-generated video and the limits of detection and watermarking also came up (PolitiFact, Eurasia Review, Live Science). Copyright and culture were raised in Africa and in Japan (Mail & Guardian, YouTube). 【Bias, discrimination and fairness】 - The ILO is putting workers' rights under the spotlight, under the headline "Who Gets a Say When AI Runs the Workplace?" (Devdiscourse) - The headline gives no measurement method or scale, so who is disadvantaged is not specified.

Privacy and surveillance

- AI chatbot conversations may not stay private in court (Mezha.net). The headline does not say which data or which jurisdiction. - Some people are falling for Meta's Muse AI (Fox News). The headline does not say what data is involved or how consent works. - A page titled "best ai chatbot boyfriend with consent and age verification" is listed under Japan's Hokuriku Regional Development Bureau, an unusual source for this topic (Japan's Hokuriku Regional Development Bureau, as listed). 【Transparency and accountability】 - EASA published a guide on AI transparency in advertising (Frankfurt Kurnit Klein & Selz, via Passle). The headline does not give its contents, so what is and is not disclosed is unknown. 【Creators, copyright and consent】 - A Mail & Guardian headline says "Do not let AI steal our culture." The headline names no specific parties or dispute (Mail & Guardian).

Disinformation and deepfakes

- Confirmed in the headlines: - Scientists can now "watermark" AI-generated proteins to deter bioweapon creation (Live Science). - PolitiFact reports that a Russian group targeted Talarico with a fake AI-generated video (PolitiFact). - Open questions and arguments: - Live Science asks whether the protein watermarking will really work. Its effectiveness is not established in the headline (Live Science). - An Eurasia Review op-ed argues that AI detection misses things in wartime misinformation (Eurasia Review). 【Protecting children and vulnerable people】 - A class action lawsuit accuses OpenAI of unlawfully extracting children's data. This is an allegation, and the plaintiffs are the ones acting (Top Class Actions). - A report finds AI boosts efficiency but fuels workplace mental health strain. Workers are the group at risk, and the headline names no one who has acted (KSNV). - A psychiatrist describes the risks of turning to AI for mental health advice (The Independent). 【Ethical principles and guidelines】 - Proposal: experts warn that public trust in government AI must start with accountability (THINK Digital Partners). - Proposal: labour standards can shape ethical AI at work (Mirage News). - Company account: Microsoft describes lessons from scaling AI responsibly in its "Customer Zero" journey. This is the company's own description (Microsoft). - None of the headlines reports a binding decision.

AI ethics in Japan

- A YouTube episode discusses a sense of copyright in the generative AI era, with a professor from Dokkyo University's Faculty of Law as guest, in a JASRAC-linked series (YouTube). - An explainer on the ethics of "AI girlfriends" is listed under Japan's Hokuriku Regional Development Bureau (Japan's Hokuriku Regional Development Bureau, as listed). - Yahoo! News comment roundups highlight security risks of generative AI and creators' conflict over whether to use AI. They describe the issues as both technical and ethical (Yahoo! News).

Statements by key figures

No notable developments. 【Integrated analysis: how the ethical questions connect】 - The headlines share one structure: the people affected by AI appear to have little say in how it is used. - Workers are the subject of the ILO question about who gets a say (Devdiscourse). - Children's data is the subject of the alleged unlawful extraction (Top Class Actions). - Chatbot users face the possibility that conversations are not private in court (Mezha.net). - Creators are the subject of the culture concern (Mail & Guardian). - A second shared structure is limited visibility into what AI does. The headlines cover advertising transparency guidance (EASA), doubts about protein watermarking (Live Science) and gaps in detection (Eurasia Review). Together they suggest disclosure and verification tools are still unproven. - Inference: employers may need to involve workers earlier when deploying AI. Companies handling minors' data may face closer scrutiny of their consent practices. Users may reasonably treat chatbot conversations as potentially discoverable. Organizations may find that public-sector accountability is a precondition for trust, as the experts quoted by THINK Digital Partners argue.

What to watch

Whether the OpenAI class action over children's data moves forward, and what it alleges about consent (Top Class Actions). 2. Whether protein watermarking is shown to work, given the doubts raised (Live Science). 3. Whether the discussion of workers' rights at the ILO leads to concrete labour standards for AI at work (Devdiscourse, Mirage News). This is an inference.