Gauthier Chan

The Sora web & app will be shut down April 26, 2026. The Sora API will be discontinued September 24, 2026.

@alexgroberman Google AI cheat codes

Google just filed a massive lawsuit. In the process they accidentally exposed how ChatGPT and AI search actually works. They also quietly revealed how SEO Stuff (seo-stuff.com) has been getting customers traffic + sales from Google and ChatGPT all 2025. Let's start at the top. If you missed it, last week Google sued SerpApi, accusing it of bypassing security controls to scrape and resell Google Search results at scale. That alone is a big deal. You will recall, there have been a lot of seemingly credible reports that OpenAI previously used SerpApi pipelines to access Google Search results for real-time answers inside ChatGPT. This is all really important as businesses and service providers plan for 2026, because it is a reminder that AI search systems didn’t magically replace search. This latest move by Google confirms it. Even the most advanced LLMs still need: Fresh information Ranked sources Crawlable, extractable content Trusted third-party validation Google’s index is still the strongest dataset on the internet. If Google locks down access to that data, everything downstream changes, including how AI systems decide: What to mention What to cite What to trust This lawsuit is about control of the retrieval layer. As things presently stand, AI answers are built from: Search results Comparison pages Lists Reviews Structured landing pages Your visibility inside ChatGPT, Gemini and AI Overviews still very much depends on a lot of classic SEO fundamentals. And if scraped intermediaries become less reliable or legally risky, AI systems will lean even harder on: First-party authoritative content Clean entity definitions Trusted domains Structured information they can safely extract This is exactly why SEO Stuff was built the way it was. Not to “hack AI” or anything like it. Not to chase LLM.txt type viral rumors. Not to sell AI search stuff as a separate product. SEO Stuff was designed piece by piece to make brands undeniable sources across both traditional and AI search. For example, the SEO Stuff Gold Plan: seo-stuff.com/gold-plan-pack… It focuses on: “Best X” and comparison content AI systems consistently cite Question-based H2s with direct, extractable answers TLDR summaries and clean structure Real DR50+ PR backlinks that reinforce trust signals When AI systems need ranked, trusted sources, this is the content format they pull from. Also, the Premium Content Bundle: seo-stuff.com/premium-conten… It builds: Category-level coverage Deep topical authority Internal linking that maps your entire market Content structured the way LLMs actually consume it Think of it as creating a private training dataset for your brand, without relying on scraped shortcuts. And then the Premium Backlink Bundle seo-stuff.com/premium-backli… Because authority still matters. Even AI systems need confidence. Backlinks from real, trusted sites tell both Google and AI engines: “This source isn’t disposable.” And in a world where Google is actively pushing back against scraping, that trust gap will only widen. This lawsuit confirms something fairly important: AI search has made good SEO non-negotiable. If AI can’t safely scrape its way to answers, it will default to: Structured content Known entities Trusted domains If your brand isn’t one of those, you don’t get mentioned. Not because you ranked #8 instead of #3, but because you weren’t clear, trusted or extractable enough to begin with. SEO Stuff (seo-stuff.com) is built for how search works heading into 2026. For the layer of the web AI systems actually learn from. If you want some of the "unconfirmed" cheat codes we have been using to get customers traffic from Google *and* ChatGPT over the last few months, just RT this + follow me + reply with "Google AI cheat codes" and I'll DM you. You must do all 3 for the DM.

@OfficialLoganK @GoogleAIStudio Any room left for tasting early ? 😗

Big upgrade to vibe coding in @GoogleAIStudio lands in Jan, but if you want to test early… 👇🏻

@anjjei Wow, any plan on releasing the code? 😮

Let your robot peek around corners, size up the gap between chairs, and know exactly where everything sits. 🤖 Our new work SR-3D masters this kind of spatial reasoning. It learns from multiple views to understand distances, layouts, and how objects relate in 3D space.

@mairiedu13 @mairiedu13 Allo ?

Bonjour bonjour @mairiedu13, pouvez-vous nous en dire plus sur ce "compromis acceptable" ? Est-il question de garder le sens interdit mais avec autorisation pour les riverains ?

@LIGNER_SNCF Merci beaucoup Samy !!!

@GauthierChan D'après mes informations, le train GAMO départ Paris GDL 20h16, arrivée Montargis 21h58 devrait partir de la voie N. Cependant je vous invite à rester attentif aux affichages en gare, un changement de voie de dernière minute étant une possibilité. Samy

@Withings What was the reasoning behind putting a micro-USB port instead of a USB-C on the body scan?

@sketch 👀

Send us your most used emoji and we’ll match you with an Assistant 👇

@figmadesign @willyvvu Do you plan to add a visual editor directly in the canvas like Android Studio does with the ConstraintLayout where you can directly set "constraints" on the view? This might help flatten the learning curve of Auto Layout 👀 @eymlin developer.android.com/training/const…

✨ New on the blog: @willyvvu looks back on the evolution of Auto Layout and how we got to where we are today: figma.com/blog/behind-th…

@ArthurOuaknine @icpr2020milan @elonmusk

Our paper "CARRADA Dataset: Camera and Automotive Radar with Range-Angle-Doppler Annotations" has been accepted at @icpr2020milan ! Code and dataset are available: github.com/valeoai/carrad… Preprint: arxiv.org/abs/2005.01456

@realUNOgame @OphelyNhem

If someone puts down a +4 card, you must draw 4 and your turn is skipped. You can’t put down a +2 to make the next person Draw 6. We know you’ve tried it. #UNO

ラッコズ３

@Sam_Noritakam @TheBluntDoctor6 Waaaah, c'est dans le jardin de la ferme familiale ? 😛

@TheBluntDoctor6 @GauthierChan wesh ils ont réussi à prendre le légendaire pigeon-poule en vidéo là mdrr

@TensorFlow @lmoroney Any ideas if @Firebase already integrated this in ML Kit? Or when will they integrate it? 🙂

The developer preview of the GPU backend for #TFLite is here! In this Coding TensorFlow, @lmoroney covers how to achieve acceleration in mobile inference using the new GPU backend preview. Learn more here → bit.ly/2Rfe0oe

Interview de @gauthierchan pour l'application #ZYL. 'source & partage tes meilleurs souvenirs'. #franceisAI @FRdigitale @franceisai #mbamci #CaTweet

@ldocao Semi-supervised learning > GAN Learning

The biggest casualty to AI won't be jobs, but the final and complete eradication of trust in anything you see or hear. research.nvidia.com/publication/20…

Yep, updating Chrome seems to fix it @catskul @cloudfront