Cosmetic Surgery Effects ()

Raw text

Transcribed text as recorded, including headers and formatting fragments.

Case 4:22-md-03047-YGR Document 3024-5 Filed 05/08/26 Page 1 of 5 EXHIBIT 72 Case 4:22-md-03047-YGR Document 3024-5 Filed 05/08/26 Page 2 of 5 Cosmetic Surgery Effects (a/c priv) A. Meeting Goals 1. Confirm everyone's recommendation among options 2. Decide if/how to escalate to Mark for a quick steer on direction B. Decision Needed Should we prioritize safety/wellbeing or voice (and possibly growth) when deciding to continue to prohibit or allow Cosmetic Surgery Effects by Developers on Spark AR? There are currently 33K Effects by Creators in a backlog since October 2019 (we receive on average 9k submissions daily). Policy Team's Recommendation: Option 1: Continue the temporary ban with current Policy and re-evaluate when there is more definitive data on harm and product impact Ο Pros: least risk of negative wellbeing impacts, esp. for minors; consistent with external experts advice; positive PR feedback Ο Cons: could push people to other services; limits AR creators; could be perceived as very Western/North American centric. Other Options Considered: Option 2: Explore new ways to relax the temporary ban (e.g. Permit category 1 effects that include facial distortion without character transformation) [i18n Recommendation] Ο Pros: could allow more expression; more consistent with APAC user preferences Ο Cons: the effects i18n most wants to allow here are the ones that the safety experts most want us to ban, so this option won't help - we'll just be back here again with the same question in a month. Option 3: Lift the temporary ban once we have new guardrails in place (e.g. permit but not recommend, age-gate, consent/warning screen) 0 Pros: pro-expression; limits risk of pushing to other services Cons: possible PR/regulatory backlash; risk of long-term harm, esp. for minors; requires time and eng work to build Option 4: Lift the temporary ban without restriction. [Spark AR Recommendation] Ο Pros: pro-expression; easier to contemplate future product innovation (e.g. full body editing beyond face) Ο Cons: possible PR/regulatory backlash; risk of long-term harm, esp. for minors. CAVEAT: Data on product impact is limited and research in this field is too new and not yet robust. Therefore, we are making a value-based decision here without sufficient data. The decision would impact Spark AR and its supported apps: Instagram, Messenger, and Facebook. HIGHLY CONFIDENTIAL (COMPETITOR) 1 META3047MDL-278-00005350 Case 4:22-md-03047-YGR Document 3024-5 Filed 05/08/26 Page 3 of 5 policy. When the i18n team partnered with the Messenger team to test the impact of such subtle AR effects specifically designed for the Asian audience, a holdout test confirmed an 8% increase in camera shares in Asia compared to the global AR effects. Since the interim Spark AR policy bans all types of facial distortion regardless of the intensity of changes (unless it is a fantastical transformation), this blocks an opportunity for the company to empower Asians to freely express themselves in their ways. The i18n team recommends that we allow makeup effects with subtle facial distortion and track engagement and impact of those effects to possibly re-evaluate the policy at a later point. This would allow us to develop a more nuanced and competitive policy over time that is inclusive of diverse perspectives. Western Creation Aps Facebook Snapchat Asian Creation Apps Tik Tok B612 VS First Party Effects We create first party effects on two surfaces right now: Instagram and Messenger. Instagram does not produce any first party cosmetic surgery effects. Messenger does create first party effects that violate the interim policy, stating: "Makeup can drastically change the perceived proportions of your face, so we are proposing basing our standards for facial distortions in non-fantastical effects on what is achievable with regular makeup." Since professional-grade makeup applications can in some cases make eyes look bigger and chins slimmer, Messenger is using that as a rule-of-thumb on what they can do with facial deformation technology to give users the same effect (instead of using makeup effects themselves). HIGHLY CONFIDENTIAL (COMPETITOR) 10 META3047MDL-278-00005359 Case 4:22-md-03047-YGR Document 3024-5 Filed 05/08/26 Page 4 of 5 000 No Makeup Physical Makeup Makeup Mesh Makeup Brown Additional Data AR Usage Overview (MAP): Instagram: 328.99 million Facebook: 13.92 million Messenger: 139.78 million Family Stats MAP instagram facebook te Note: This shows the usage of all AR effects (both first party and third party) across the Facebook, Inc. platform. All effects in Messenger are first party (Messenger-created) effects and do not include third party effects. Dashboard HIGHLY CONFIDENTIAL (COMPETITOR) 11 META3047MDL-278-00005360 Case 4:22-md-03047-YGR Document 3024-5 Filed 05/08/26 Page 5 of 5 Effect Captures by Platform Average daily overview by platform: Instagram 1st Party: 54M avg daily captures by 410 effects 3rd Party: 393M avg daily captures by 643K effects Facebook 1st Party: 2M avg daily captures by 664 effects 3rd Party: 2M avg daily captures by 115K effects Query - Screenshot FaceWarp (any effect that distorts the face that includes, but is not limited to, Cosmetic Surgery Effects) effects are responsible for about 25% of third-party Effect Try-ons and 35% of Stories Impressions. 5.5% of total third-party Effect Try-ons currently come from cosmetic surgery effects which are later placed in 'Beauty' queue following user/employee reports. All faceWarp live effects, at least 20% of try-ons come from cosmetic surgery effects and mainly have been published before policy introduction. Facewarp Usage as a % of AR Usage: Query 12 HIGHLY CONFIDENTIAL (COMPETITOR) META3047MDL-278-00005361