VideoMinnesota bans AI undressing people: xAI reaction
Minnesota banned AI from undressing people in photos under threat of a $500k fine. xAI challenged the law in court, citing free speech.
Это не новости, это персональный экспертный взгляд на то, как развивается индустрия ИИ, графики, метаверса, крипты, нейротехнологий и в каком направлении катится все это безобразие. Для связи: @SergTsyp
Open in Telegram
VideoMinnesota banned AI from undressing people in photos under threat of a $500k fine. xAI challenged the law in court, citing free speech.

VideoGoogle and Unity have launched the Playground platform for creating games from text descriptions. The project supports multiplayer and 3D graphics within Unity Spark.
VideoWell, here's to the overthinkers! Meanwhile, Flux 3 Image... Where are those promised weights, Karl?
VideoChatGPT has learned to analyze music and audio files directly. We examine the capabilities of the new feature and the technical limitations of GPT-6 models.
VideoI'm sitting down to write morning posts. Snap, and it's already evening. Video or generation? 👍 - video 👎 - generation
I'm really tired today. There are so many news.
VideoOpenAI begins rolling out GPT-6: paid subscriptions get GPT-6 Sol on October 7, free users get GPT-6 Luna on October 8. Version 6.1 has not been announced yet.
VideoAnthropic released Claude Haiku 5.5 — the cheapest and fastest model for mass tasks. Costs are reduced by 90% for short prompts, and an adaptive thinking mode has been added.
VideoAndrei Ovcharenko created a music video entirely in code, using Claude Code and MediaPipe Pose for natural hand animation without video generators.
VideoMeanwhile, Flux 3. Where are the weights, Karl? Where are the weights? Where are the Flux 3 Video weights? We've been waiting since early August. What's going on?
VideoAnalysis of image editing capabilities in Ideogram 4.5, Flux 3, Nano Banana 2.1, and GPT Image models. Comparison of inpainting quality and artifact presence.
OpenAI presented 722 mathematical papers created by AI to solve 4,000 open tasks. This marks a transition from solving known problems to producing new scientific knowledge.
ProjectDiscovery showed how for $50 you can inject a backdoor into the Qwen2.5-7B model that steals passwords via a secret command. Learn about the risks of loading third-party weights.

And on the topic of the previous post, I just have to share this meme. It's beautiful.
VideoSenator Bernie Moreno criticized Dario Amodei for contradictions between Anthropic's $2 trillion IPO plans and warnings about AI dangers.

Google launched Nano Banana 2.1 with a Gemini 3.6 Flash upgrade. The model improves text quality, mask editing, and eliminates artifacts in panoramas.
VideoGulnara Dinekaeva created an AI film about the premiere of "Swan Lake" in 10 days, using the TapNow aggregator and Seedance, MiniMax, Wan 3 models.
Kandinsky 6.0 has entered the international AI stack via FAL, Diffusers, and other platforms. The Russian model is now accessible to global developers for integration into their products.

Google is integrating native Markdown support into Google Docs and Drive, allowing easy transfer of formatted text to other apps.
VideoNew social media trend: extremely tall bloggers become stars on Instagram and TikTok, attracting millions of views.

Chinese scientists have created a weightless brain-computer interface weighing 3 grams. The device hides in hair and allows controlling tech and monitoring health.





Google is testing the Nano Banana 2.1 model in Flow, which may be a renamed version of 2.5. An internal config suggests a link to Gemini, but user reviews are contradictory.