AI Ethics News — 2026-11-07 (Sat)

Overview

Several ethical questions surfaced in the past 48 hours. One is consent and disclosure in surveillance: Hong Kong police are adding AI facial recognition to CCTV, and police say the plan complies with privacy law (South China Morning Post, Hong Kong Free Press). A second is whether AI reproduces human bias (Bioengineer.org, GLAAD). A third is disclosure of synthetic content and compensation for content creators (Arab News, Marketing Brew). 【Bias, discrimination and fairness】 - A study reports that AI reasoning processes consistently reproduce human stereotypes (Bioengineer.org). The headline gives no measurement method or sample size. - GLAAD published its 2026 AI report on LGBTQ impacts across AI (GLAAD). The headline does not say what was measured. - An employment law outlet covers cases where the use of AI conflicts with employees' sincerely held religious beliefs in the US (Employment Law Worldview). Employees are the group affected.

Privacy and surveillance

- Hong Kong police are to introduce AI facial recognition into the Smart View surveillance network (The Standard (HK)). They are to enable it on the CCTV network this month (South China Morning Post). - Police say that linking mall cameras to the AI facial recognition network complies with Hong Kong privacy laws (Hong Kong Free Press). That is the police position. The headlines do not say whether the people filmed have consented, or what the data will be used for beyond surveillance. 【Transparency and accountability】 - Saudi Arabia has set out AI principles for media and calls for disclosure of synthetic content (Arab News). - Another report says AI disclosure and privacy are at the center of the journalism debate (Arab News). - The headlines do not say how disclosure would be enforced. 【Creators, copyright and consent】 - Content compensation in the agentic era is being pushed (Marketing Brew). The parties and details are not given in the headline. - Nine's boss called a mooted AI licensing model "deeply flawed" (Michael West Media). The dispute is over the terms on which creators' content is licensed for AI.

Misinformation and deepfakes

- More than 100 charities urge the government to address misinformation in AI-generated health information (Third Sector). This is a call from the charities, not a confirmed finding of harm. - A headline reports that South Korea ranks 2nd among 10 countries in AI deepfake sex crime victimization, with a victim saying "Nobody cared, nobody believed me" (Herald Business, Huam-dong Research Lab). - On countermeasures, a guide covers what to label under Instagram's AI disclosure policy (Hootsuite Blog). 【Protection of children and vulnerable people】 - The Electronic Frontier Foundation (EFF) submitted comments to a United Nations human rights office (OHCHR) consultation on child rights and safety in the digital environment (EFF). The headline does not give the content of the submission. - Victims of deepfake sex crimes in South Korea are also a vulnerable group (Herald Business). 【Ethical principles and guidelines】 - Decided: Saudi Arabia has set out AI principles for media (Arab News). - Launched: UNESCO has launched the second edition of the "Abu Rayhan Beruni" International Prize for AI Ethics (Dunyo). - Educational practice: Pathways World School Gurgaon is described as setting a new direction for responsible AI in education (Business Standard). - The EFF also reflects on digital rights wins and losses in Sacramento (EFF).

AI ethics in Japan

- Toyo Keizai Online asks why business people should question AI ethics now, and what to entrust to AI and what to leave to people, for the sake of a good society (Toyo Keizai). - E-Guardian has begun a new service that monitors misuse of characters by generative AI (kabu-ir.com). - Golkar of Indonesia aims at digital transformation and is preparing AI ethics guidelines (VOI.ID). This item is not about Japan.

Statements by key figures

No notable developments. 【Integrated analysis: what is happening】 - Missing consent appears across sections. Facial recognition on CCTV raises the question of whether the people filmed have agreed (Hong Kong Free Press). Compensation for content likewise raises whether creators agreed to use of their work (Marketing Brew, Michael West Media). - Missing disclosure is the second shared structure. Saudi Arabia calls for disclosure of synthetic content (Arab News), and Instagram has a labeling policy (Hootsuite Blog). The deepfake victim's remark suggests that harms may go unrecognized when sources are not visible. - For companies, this likely means labeling AI-generated content, documenting the legal basis for surveillance, and checking whether employees' beliefs are considered. These points are inference, not content of the headlines.

What to watch

- Whether Hong Kong police's facial recognition launch this month is accompanied by disclosure of its uses and safeguards. This may show how a "complies with privacy law" claim is tested. - Whether Saudi Arabia's synthetic content disclosure principles are followed by concrete labeling requirements. - Whether the AI licensing models criticized by Nine's boss lead to terms creators accept.