Skip to content

Tempo Detektif Deepfake

FreeFirst-line

Operator: tempo.co | Type: Non-detector | Cost: Free

TL;DR

A Bahasa-Indonesian inoculation newsgame from Tempo Cek Fakta that walks players through deepfake-recognition scenarios using a multiple-choice format. WAN-IFRA Digital Media Awards Asia 2026 Silver in Best in Countering Disinformation. The fact-checker's deployment tool, not the fact-checker's verification tool; its purpose is to inoculate audiences before they encounter the synthetic content the fact-checker is debunking.

What it does

Detektif Deepfake is a browser-based newsgame that presents players with deepfake-recognition scenarios in Bahasa Indonesia, asking the player to identify manipulation cues (visual artefacts, audio mismatches, contextual signals) through a multiple-choice interface. The pedagogical foundation is inoculation theory: by exposing players to weakened forms of manipulation techniques in a low-stakes setting, the game builds recognition skills that transfer to real encounters with synthetic content in the wild.

The game was developed by Tempo Cek Fakta (Tempo's fact-check desk) and Tempo Media Lab as part of the IFCN ENGAGE programme and launched in May 2025. It won the WAN-IFRA Digital Media Awards Asia 2026 Silver for Best in Countering Disinformation in Manila on 28 April 2026. The lead author is Inge Klara Safitri. Two companion games (Vaksin HOX for general hoax inoculation and Anti Jebakan Batman for phishing) extend the same pedagogical approach across adjacent disinformation surfaces. Detektif Deepfake itself recommends Deepware.ai and MIT Media Lab DetectFakes inside the game's "go further" guidance.

For the fact-checker, the game is something to deploy outward rather than something to apply inward. The fact-checker who embeds Detektif Deepfake on the Tempo Cek Fakta page, shares it through a partner newsroom's social media, or hosts it inside a school workshop is using the tool the way it is designed to be used.

When to use it

  • A regional newsroom is preparing audiences for an election cycle and wants a Bahasa-Indonesian inoculation surface that reaches readers before deepfakes circulate widely.
  • A media-literacy programme in Indonesia is training secondary- school or university students and needs an interactive, Bahasa-native game that teaches deepfake-recognition through practice rather than lecture.
  • A community-volunteer training session needs a short, engaging warm-up that teaches the manipulation cues fact-checkers themselves rely on when triaging suspect video.
  • A coalition campaign (CekFakta, MAFINDO, Tempo combined) is preparing audiences in advance of an anticipated synthetic- political-content surge: election season, a foreign-policy flashpoint, a known impersonation actor.

Limitations

Limitations

  • Single-format game. The multiple-choice newsgame format is one inoculation surface; effects typically require booster sessions to maintain over months rather than days.
  • Should be ported to other SEA languages (override flag should-port-other-SEA-languages-pointer). Bahasa Indonesia coverage only at present; Thai, Filipino, Sinhala, Tamil, Malay, and Lao audiences cannot use the game in their working language. The toolkit's cell-level honest-gap names the absence of localised prebunking effectiveness measurements outside Indonesia.
  • Inoculation effects measured for prebunking generally decay over weeks without reinforcement; the literature base for Bad News and Jigsaw Prebunking puts the half- life in the order of weeks, and Detektif Deepfake inherits that shape rather than overcoming it.
  • The game teaches recognition of manipulation cues that newer synthetic-content generators may not produce; the game's effectiveness against the most current deepfake generation methods is a function of how recently the game's scenarios were updated relative to the generator landscape.

Privacy and threat model

The game is a publicly accessible web page. Players load it in the browser; no account creation is required for casual play. Tempo's standard web analytics apply to the page; for any deployment that integrates the game into a newsroom workflow beyond casual play (curriculum integration, formal training, community-volunteer programmes) the operator should review Tempo's data-handling terms before deployment.

The threat model that matters here is the deployment model rather than the data-flow model; a fact-checker embedding the game does so for an audience the fact-checker has identified as worth reaching. The game itself does not handle source- identifying material because it works on pre-built scenarios, not on user-submitted content.

Country and platform applicability

  • Indonesia: primary deployment. Tempo Cek Fakta uses the game in audience-engagement and education programming; the WAN-IFRA Silver context establishes regional editorial recognition.
  • Laos: not applicable in the source language; Bahasa Indonesia content is not accessible to Lao audiences. The cell-level honest-gap on localised prebunking outside Indonesia applies.
  • Malaysia: Bahasa-Malaysia speakers can read the Bahasa-Indonesia content with comprehension, but the game is not adapted to Malaysian context, references, or named entities. Practical reach is limited; the inoculation case for Malaysia would benefit from a Malay adaptation.
  • Philippines: not applicable in the source language; the underlying inoculation pattern transfers but a Filipino adaptation does not yet exist .
  • Sri Lanka: not applicable in the source language; Sinhala or Tamil adaptations would require local Tempo- equivalent partners to lead.
  • Thailand: not applicable in the source language; a Thai adaptation could plausibly be led by Cofact or AFP Fact Check Thailand if regional partnership emerged.

Platform applicability: web-deployed; the game can be embedded or linked from any platform a newsroom or training operator uses to reach their audience (Facebook, Facebook Groups, TikTok organic posts, LINE, WhatsApp shareable links).

How to access

Free public access through the Tempo Cek Fakta website. The game is one of three companion newsgames Tempo publishes (alongside Vaksin HOX and Anti Jebakan Batman); Tempo's fact-check section is the entry point. No account required for play; institutional integration (embedding into curriculum, deploying through partner newsroom social media) is a partner arrangement with Tempo rather than a configuration step.

Cost (current as of 2026-05)

Free for players and for institutions integrating the game into audience-engagement work. Tempo's institutional partnership costs sit outside the player-side product; the game itself is published under a free editorial licence.

Quickstart (deployment-side)

(The user reading this card is a fact-checker deploying the game to audiences, not a player verifying suspect content. The quickstart reflects that.)

  1. Identify the audience moment: an election cycle on the horizon, a media-literacy curriculum slot, a community- volunteer training session, an anticipated synthetic-content surge.
  2. Visit the Tempo Cek Fakta page and locate Detektif Deepfake; confirm the game's current scenarios are recent enough for the audience moment (the multiple-choice scenarios reflect the synthetic-content landscape they were built against).
  3. For audience deployment through a partner newsroom or NGO, share the game's URL through that organisation's social- media channels alongside framing copy that names the inoculation purpose ("five minutes of practice that makes spotting deepfakes easier").
  4. For workshop deployment, plan a 15-30 minute session that includes the game plus discussion of the manipulation cues the player encountered.
  5. For curriculum integration, partner with Tempo and adapt the game into the broader media-literacy curriculum frame alongside Vaksin HOX and Anti Jebakan Batman.
  6. After deployment, plan reinforcement: booster sessions, follow-up content, or pairing with Bahasa fact-check coverage; inoculation effects decay without reinforcement.

In the toolkit's workflow

Source-history and behaviour pillar non-detector audience- engagement deployment per Architectural Anchor 1. Sits in 2A.3 as the primary entry: the SEA-grounded gold-standard inoculation game that the toolkit privileges as the default reference for any prebunking deployment in or about Indonesia.

Standard combinations (deployment combinations):

  • With IREX L2D / Gali Fakta at 2A.3 – IREX's curriculum-style media-literacy training pairs naturally with Detektif Deepfake's interactive game format; the two together form a complete media-literacy programme for Bahasa audiences.
  • With Jigsaw Prebunking at 2A.3 when a campaign needs both a deeper interactive game (Detektif Deepfake) and shorter video-ad inoculation surfaces (Jigsaw) reaching passive audiences on YouTube, Meta, and TikTok.
  • With Bad News / Harmony Square at 2A.3 when a trainer wants to teach the broader inoculation pattern alongside the deepfake-specific game; Bad News covers the six general manipulation techniques, Detektif Deepfake covers the synthetic-content surface.
  • With Kalimasada and Yudistira at 2B – the prebunking work happens upstream of the tipline intake and downstream feedback (community questions about scenarios) can flow back into the MAFINDO/CekFakta workflow.

Decision-tree references: this is not a verification routine; the decision trees in T1–T7 do not invoke Detektif Deepfake at runtime. The tool sits in the editorial-strategic layer that shapes audience preparation rather than per-claim handling.

This card carries no detector signal class: Detektif Deepfake is an inoculation deployment that builds audience recognition; it does not produce a synthetic-content verdict on any specific piece of suspect content.

Override notes

Override notes

  • WAN-IFRA Silver April 2026 (wan-ifra-silver-apr-2026- pointer): the game won the WAN-IFRA Digital Media Awards Asia 2026 Silver for Best in Countering Disinformation in Manila on 28 April 2026. The award context anchors the regional editorial recognition that justifies Detektif Deepfake's primary status in 2A.3.
  • Should be ported to other SEA languages (should-port- other-SEA-languages-pointer): Bahasa Indonesia coverage only at present. The cell-level honest-gap on localised prebunking effectiveness outside Indonesia names the structural absence of Thai, Filipino, Sinhala, Tamil, Malay, and Lao adaptations; partner-led adaptation is the operational route to closing the gap.

Sources

  • Safitri, I.K. et al. Detektif Deepfake — deepfake literacy newsgame. Tempo Cek Fakta / Tempo Media Lab, May 2025. tempo.co/tag/detektif-deepfake. (WAN-IFRA Digital Media Awards Asia 2026 Silver, 28 April 2026, Manila — cited in card)