Documentation
Notes for players, with links to primary sources including the ASEAN Guide on AI Governance and Ethics and ISO/IEC AI standards. Sources are listed at the bottom of this page. This is not legal advice.
AI risk
You just tried to tell a real photograph from a generated one. That skill is useful, and it is not a control. The ASEAN Guide on AI Governance and Ethics (2024) treats generative systems as a distinct class of risk: they can invent text, images, and audio that look human-made, then scale mistakes, deepfakes, and disinformation. This page is the point of the game.
The Guide is built so that design, development, and deployment consider broader social impact, not only model quality.
Transparency and explainability
People should be told when AI is in the loop, what data it uses, and why a result appeared, so they can choose whether to trust it.
Fairness and equity
Outputs must not amplify discrimination. Training data and checks have to represent the people who will be affected.
Security and safety
Known harms should be assessed and mitigated, with a way for a human to stop an unsafe decision.
Human-centricity
Systems should serve human well-being, not replace judgement where the cost of a fake is high.
Privacy and data governance
Personal and sensitive data used to train or prompt a model must be collected and held under clear rules.
Accountability and integrity
Someone has to own the outcome: who built it, who deployed it, and who answers when it is misused.
Robustness and reliability
The same situation should not produce a convincing lie one day and a different lie the next.
ASEAN’s regional recommendations say governance should include guidance on how to distinguish AI-generated content from authentic content. One path it names is a digital watermark on synthetic files, so people know they are looking at a generation and can decide with that fact in hand. This quiz trains attention. Watermarks, labels, and provenance are the actual defences.
Modern image models can invent photorealistic scenes that never happened. People often score only a little above chance on tests like this, and extra editing can hide the remaining tells. Treating “I can tell” as a security control is the risk: it fails quietly, at scale, in a feed you scroll past.
ISO/IEC 42001 is the international standard for an AI management system. If an organisation develops, sells, or uses image generators, the concern is not whether a sharp-eyed colleague can tell a generated picture from a photograph. It is whether someone owns the harms: misuse, deepfakes, fraud, and the collapse of trust in genuine evidence. The standard expects policy, risk assessment, and treatment — not a person squinting at pixels. ISO/IEC 23894 is the companion guidance for running that AI-specific risk process. For the files themselves, ISO/IEC 21617 (JPEG Trust) is the media-authenticity track: provenance, integrity, and attribution annotations so trust sits in the asset, not in the viewer. This quiz is literacy. It is not an ISO/IEC control.
ASEAN names deepfakes, impersonation, and malicious activity as generative-AI risks: images and video that convincingly show people doing things they never did, used to extract information or to impersonate someone you think you know. The same techniques that made a convincing landscape can impersonate a person, a product, or a news event. A second-order harm is the liar’s dividend: once fakes are common, genuine evidence can be waved away as AI.
ASEAN’s first guiding principle is transparency: tell people when an AI system is being used, what it is for, and that they are not dealing with a human if they are not. For products, that means a general disclosure when AI is in the offering. For generated media, the EU AI Act’s transparency rules (from August 2026) require machine-readable marking of synthetic audio, image, video, or text, and visible disclosure for certain deepfakes. Labels can still be stripped by a screenshot. Pause anyway.
ASEAN tells governments to raise public awareness so citizens can protect themselves from AI used to generate or spread misinformation. Media literacy: identify and verify a source you already trust, not only the pixels. Check whether the file carries Content Credentials. For anything that would move money, votes, or someone’s reputation, assume a still image can be fabricated and verify through a second channel. This quiz does not prove a picture is real.
