What changed for people posting AI video from Brazil?
On 5 November 2026 the Google blog introduced Gemini Omni 1.2 Flash. It says the outputs carry an imperceptible SynthID watermark, and it describes native clips of up to 15 seconds, with a length you set in steps of 1 second from a minimum of 3 seconds. The same post mentions directing dialogue between several characters in Flow. The figure of US$ 0.07 per second for 720p on that page is a dollar price, not the Flow credit table.
On the Google DeepMind page, SynthID is an invisible digital watermark placed in AI-generated images, audio, text or video. For images and video, that page says the mark is added when the content is created and is designed to stand up to cropping, filters, frame-rate changes and lossy compression. The same page says the SynthID Detector portal is now open to everyone and checks whether a file was made with AI from Google or its partners. The Google blog updated on 29 October 2026 says the tool is available to everyone, globally, in English, and that it checks content from Google and from partners it names, including OpenAI, NVIDIA, Kakao and Apple. The same post says Google has watermarked over 180 billion images and videos and 240,000 years of audio, and that checks in Search, the Gemini app and Chrome handle over 1 million requests a day. DeepMind also says you can upload a file in Gemini, Search or Chrome and ask whether Google AI created or altered it.
That hidden mark is a different thing from the label a follower sees on Reels, TikTok or YouTube. The on-screen label is a step you take when you publish.
The TikTok Shop page for Brazil prints the date 11/09/2026. It says AI content is allowed when it is honest, and that creators must disclose content that was fully generated or significantly edited with AI. Disclosure matters especially when the video or audio has a synthetic face, a voice, a digital human or a highly realistic virtual figure, when the edit changes the meaning, or when a viewer could mistake the clip for real footage.
When do YouTube, Instagram and TikTok ask for the label?
The lines below come from the English pages read on 13 November 2026. This guide is not legal advice, and a later edit of those pages can change a detail.
| Place | When the page asks for a label | How the page says to add it | What the page says if you skip it |
|---|---|---|---|
| YouTube | Realistic content that makes a real person appear to say or do something they did not do, alters footage of a real event or place, generates a realistic scene that did not happen, or makes AI-generated music the main focus of the video. Clearly unrealistic content and minor aesthetic edits are outside that requirement. | In YouTube Studio, during upload, under Attributes, set “AI use” to Yes. The YouTube blog of 27 May 2026 says the photorealistic label sits below the player on long videos and as an overlay on Shorts. Unrealistic, animated or lightly altered disclosures stay in the expanded description. | Creators who consistently choose not to disclose can have a label applied for them, videos removed, or a suspension from the YouTube Partner Program. Disclosing alone does not limit the audience or eligibility to earn money. |
| Instagram, Facebook and Threads | Photorealistic video or realistic-sounding audio that was digitally created or altered, including with AI. This comes from Meta’s newsroom post of 6 February 2024, updated on 1 April 2025. | The post says people must use Meta’s disclose-and-label tool when they post that kind of content. It does not name the menu button. | Meta may apply penalties if people do not disclose. The post also says Meta could not yet detect those signals in video and audio made with other companies’ tools. |
| TikTok | The Community Guidelines, effective 24 September 2026, require a label on AI-generated or significantly edited content that shows realistic-looking scenes or people. | On the Brazil TikTok Shop page: on-screen text, a sticker, a note in the description, or the “AI-generated content” toggle. The page says that toggle adds a label reading “Disclosed by creator as AI-generated”. TikTok’s own AI effects are labeled for you. A clip from an outside tool, such as Flow, still needs your disclosure. | Unlabeled content may be removed, restricted or labeled, depending on the harm. On TikTok Shop, labeling a video as AI when you did not use AI is also a violation, and so is failing to disclose when AI plays a significant role. |
YouTube may add a label itself for video made with YouTube’s own generative tools, for files that contain C2PA metadata, and when its systems detect content as AI-generated or altered. The help page says you cannot clear the disclosure on YouTube’s own AI tools, on C2PA files, or after a manual review. The May 2026 blog names Veo and Dream Screen as YouTube tools in that permanent group. That sentence is about YouTube’s tools. It does not say that every file exported from Flow gets the same permanent label.
None of the platform pages read for this guide say that Instagram, TikTok or YouTube read SynthID in order to place the label. YouTube talks about its own tools, C2PA and internal detection. Meta talks about industry-standard signals such as C2PA and IPTC metadata. TikTok Shop talks about technical metadata: when that metadata is present the shop page says the platform label cannot be removed, and that the automatic label alone does not add a penalty or a distribution limit.
A label does not make a misleading shop video allowed. The Brazil TikTok Shop page says you must not use AI to change a product’s size, color, shape, features or performance so the clip no longer matches the item for sale, invent a doctor or a similar authority to endorse a health product, or show a result the product does not really deliver. Even a labeled TikTok can still break the rule against harmful or publicly misleading AI.
How many credits does a clip that needs a label spend?
Gemini Omni Flash spends Flow credits on every generation and every edit. The Flow credits help page prices Omni Flash 1.2 by the second. The totals below are arithmetic from that table, not a total Google prints for a finished clip. The page says the rates can change and that you should check Settings on the prompt box before you generate. One request can produce more than one generation, and the rate is per generation.
| Work in Flow | Who can use it | Credits per generation |
|---|---|---|
| Omni Flash 1.2, up to 10 seconds, 720p | Every account | 3 per second (10 seconds = 30) |
| Omni Flash 1.2, 15 seconds, 720p | Plus, Pro or Ultra | 3 per second up to 10 seconds, then 6 per extra second (15 seconds = 60) |
| Omni Flash 1.2, up to 10 seconds, 360p | Every account | 1 per second (10 seconds = 10) |
| Omni Flash 1.2, 15 seconds, 360p | Plus, Pro or Ultra | 1 per second up to 10 seconds, then 2 per extra second (15 seconds = 20) |
| Extend by 10 seconds at 720p | Plus, Pro or Ultra | 3 per second added (10 seconds = 30) |
| Upscale a 15-second clip to 1080p | Plus, Pro or Ultra | 0 for the first 10 seconds, then 1 per extra second (15 seconds = 5) |
| Upscale a 15-second clip to 4K | Ultra only | 50 for up to 10 seconds, then 5 per extra second (15 seconds = 75) |
A realistic shop clip is rarely one take. Ten generations of a 15-second 720p clip are 10 × 60 = 600 credits before an edit, a 9:16 crop of the same idea, or a 4K upscale. Drafting at 360p spends less while you check that the product still matches the real item. The guide on Omni Flash 1.2 credits also covers the Flow note that, on 5 December 2026, Veo 3.1 Fast, Veo 3.1 Quality and Omni Flash 1.1 are retired.
Veo 3 · Unlimited Credit means the shop refills the Family Flow pool, with no cap on the number of refills. It does not mean a generation costs zero credits. Turning on the platform label does not refund the credits the clip already spent.
Which Veo 3 · Unlimited Credit plan fits?
ChugAx is an independent shop in Vietnam. It sells access to Google AI Ultra 20x plans and is not affiliated with Google. Prices are set in VND for the whole month. Reais and dollars below are references only: 5,200 VND per BRL, and 25,800 VND per USD. The amount you pay is the one the bot confirms.
| Plan | Who shares the pool | Price for the month |
|---|---|---|
| Veo 3 · Unlimited Credit · 1 member | The shop manages the Family. Up to 5 customers in that group share one pool. | 8,000,000 VND (about US$ 310.1, about R$ 1.538,46) |
| Full Family · Unlimited Credit | You manage the Family. You and 4 people you invite share the pool. | 40,000,000 VND (about US$ 1,550.4, about R$ 7.692,31) |
If you post alone and accept a pool shared with other customers in the shop’s group, the 1 member plan is the one that matches. If your own team needs the pool and you want to send the invites, Full Family is the one that matches. Ask the bot about the support tool that helps you use a shared pool. What the warranty covers is on Is a family seat safe?
How do you order from Brazil?
Orders go through the Telegram bot @chugax_gl_bot. The website does not create the order and does not take the payment. The shop only needs your Gmail address. It never asks for a password, an OTP or a recovery code.
- Open the bot and choose the plan.
- Wait until the bot says your Gmail is eligible and states the amount and the method.
- Pay only after that message. From Brazil there is no Pix key. You buy USDT with Pix and send it on the network the bot names (TRC20 or BEP20), or you top up Wise with Pix and send VND to the MB Bank account the bot gives.
- Send the TxID or the receipt back to the bot.
The shop activates Google AI Ultra 20x within 1 to 3 days after it has the details and has confirmed payment. The steps that start with Pix are on Can I pay with Pix?
Which limits still apply?
Refills run from 10:00 to 21:00 Vietnam time (GMT+7), which is 0:00 to 11:00 in Brasília, with 2 to 3 hours between refills. Message the bot when the pool hits zero. Outside that window the refill waits until support is open again. This is not an instant, 24-hour service.
- Google and Flow limits still sit on top of the credit count. A 15-second clip and an extension need a paid Google AI plan.
- SynthID, a C2PA file or an automatic platform label does not finish the disclosure. If the clip looks real, use the label the network asks for.
- A clear “no SynthID” result does not prove a camera shot the file. The detector looks for SynthID from Google and from the partners named on the Google blog.
- The label does not authorize a fake product result, a fake expert or a real person’s likeness. YouTube’s community guidelines still apply to AI video.
- ChugAx cannot change YouTube, Meta or TikTok rules.