OpenAI just gave Sora 2 a big upgrade

The new update adds a Storyboard Tool (beta) for ChatGPT Pro users, letting you plan videos frame by frame, edit outlines, and refine scenes before rendering.

All users can now make 15-second videos, while Pro users get up to 25 seconds using storyboards.

The feature is live on web and app in the US and Canada, giving creators more control for emotional and educational storytelling, with better physics and audio sync than ever.

🎬 Google just dropped Veo 3.1 and it’s a filmmaker’s dream.

Veo 3.1 brings richer audio, more narrative control, and lifelike realism that captures texture, motion, and atmosphere like never before. It’s built on Veo 3, but with stronger prompt adherence and major upgrades in both visuals and sound.

Inside Flow, Google’s AI filmmaking tool (with over 275M videos already made!), creators now get deeper editing power:

🎧 Audio added to Ingredients to Video, Frames to Video, and Extend
🎨 “Insert” tool to add new elements naturally into scenes
🧩 Upcoming feature to remove unwanted objects or characters seamlessly

Veo 3.1 is available today in Flow, the Gemini API, Vertex AI, and the Gemini app, opening the door for the next wave of cinematic AI storytelling.

Jeremy Renner is back as the voice of the Oracle of Time in Stardust Future: Stars and Scars, billed as the first feature-length film made entirely with AI. Hitting theaters in November, the Yi Zhou–directed project blends 2D/3D animation and follows a journey from the birth of the universe to 2080.

➜ Renner reunites with Zhou (Chronicles of Disney)
➜ Part of proceeds go to the Rennervation Foundation and MPTF
➜ Awards push planned (animated feature, score, VFX, more)

📺 DirecTV is turning the idle screen into a shopping scene

➜ Starting next year on Gemini boxes, a new AI screensaver (from Glance/InMobi) can place you, family, even pets into a custom scene
➜ See look-alike outfits, furniture, accessories, tap a QR code to open the Glance app and buy on your phone
➜ Talk to your remote to tweak colors, styles, locations; there’s also a 30-sec “craft my moment” AI video mode
➜ Glance says items are matched by similarity (no single brand push) and could expand beyond the screensaver to the TV launcher

Hailuo 02 T2V prompt share

Prompt:
fast forward tracking camera movement: The camera follows closely behind a buzzing bee zipping through a lively open-air market. It weaves between fruit stands, dodges people, flies under awnings, and passes colorful stalls. Motion blurs and speed intensifies as the bee homes in on a small sweets shop. It dives through the door toward a golden honey-drizzled pastry on display. Light gleams off syrup as the bee lands, wings slowing. Vibrant, whimsical energy.

DeepSeek just released DeepSeek-OCR, a new open-source vision-language model for ultra-efficient text extraction.

The benchmarks show exceptional precision with compression rates up to 20× on the Fox dataset, and state-of-the-art performance on Omnidocbench, beating major models in both accuracy and token efficiency.

What stands out? DeepSeek-OCR handles long, text-dense documents with fewer vision tokens while keeping precision near 97%. It’s also fully open-sourced on Hugging Face and GitHub.

Wan 2.2 Prompt share

Prompt:

360-degree rotating shot: A martial artist fights off multiple enemies in a rain-soaked alley, spinning and flipping with precision strikes as lightning flashes. cinematic stylized combat footage

Generated on https://truevideo.ai/

Midjourney V7

[Prompt] --chaos 12 --ar 3:4 --exp 15 --sref 3706101356 --sw 500 --stylize 500

Generated on https://truevideo.ai/

Kling 2.5 Prompt share

Prompt:

A surfer rides a towering wave during sunset, drone camera orbiting overhead as water sprays into the glowing sky, creating a golden halo around the athlete.

Generated on https://truevideo.ai/

Your voice. Your face. Your story, now perfectly consistent across every frame.

With HeyGen + Veo 3.1, you can bring cinematic storytelling to life like never before. Choose your voice once, and HeyGen blends your real tone and emotions into Veo’s stunning cinematic motion. Whether it’s a documentary-style video, a short film, or a creative ad, your digital self stays authentically you.

No more breaks between scenes or inconsistencies in tone. Each transition flows seamlessly, keeping your identity intact from start to finish. You’re not just generating clips, you’re crafting cohesive stories where every moment connects naturally to the next.

And the best part is that your avatar becomes a versatile storyteller. The same face, the same voice, just different outfits, moods,and worlds. From casual vlogs to cinematic universes, your AI twin adapts effortlessly without ever losing that spark of “you.”

HeyGen and Veo are redefining what it means to be consistent in AI filmmaking.One identity, infinite stories.

Alibaba enters the AI wearables race.

The company just priced its Quark AI Glasses at ¥4,699 ($660) with pre-sale starting Oct 24 and shipments in December. Powered by its Qwen LLM and Quark AI Assistant, the glasses handle real-time translation, hands-free calling, and music streaming.

Alongside the hardware, Alibaba launched its AI Chat Assistant inside the Quark app, a ChatGPT-style chatbot that merges AI search and conversation in one place, powered by Qwen3 models.

The move signals Alibaba’s next focus: AI for consumers, not just enterprise clients.

Meta, Xiaomi, and now Alibaba, the battle for smart glasses just got a lot more interesting. 👓

Sora roadmap update: here’s what’s landing next

➜ Character cameos in days
➜ Cameo your dog, guinea pig, or favorite toy
➜ Create cameos from your own Sora characters
➜ Trending cameos shown live in the generation UI
➜ Basic video editing starts with clip stitching
➜ Better social with friend groups and community channels
➜ Quality of life: smarter feed, fewer false flags, faster app
➜ Android version coming soon

Meta just laid off 600 employees from its AI division, but says there’s “no cause for concern.”

Most cuts came from older, less effective teams, while Meta’s Superintelligence Labs remain untouched. The company says smaller teams mean faster progress and better impact.

Those affected can reapply for other roles, with severance packages offering 16 weeks of pay + 2 weeks per year of service.

Meta insists this isn’t about financial pressure, but streamlining ahead of massive AI investment, with expenses projected to hit $114–118B in 2025 and climb higher in 2026.

A new 1GW data center in Texas will power future AI hardware, signaling that Meta’s long game in superintelligence is still very much alive.

Hailuo 02 T2V prompt share

Prompt:
Anime-style medium shot: In a shattered school courtyard under a lavender twilight sky, two high schoolers, both with torn uniforms and glowing battle auras, stand mere inches apart. Steam rises from cracks in the concrete around them. She’s breathing hard, fists still clenched. He’s bleeding slightly, but smiling. A slow breeze moves her hair across her face. The tension breaks as he gently tucks it behind her ear. Her eyes widen. A soft lo-fi beat filters in like memory, as the world around them fades to vignette and sakura petals drift past.

Generated on https://truevideo.ai/

Reddit is suing Perplexity in New York federal court for allegedly scraping Reddit content to train its AI “answer engine.”

Reddit says Perplexity worked with third parties to bypass protections and pull data at scale, even after a cease-and-desist that was followed by a fortyfold jump in Reddit citations. The suit seeks damages and an order to block use of Reddit data.

Perplexity says it provides factual answers with accurate AI and rejects Reddit’s claims. Oxylabs and SerpApi also dispute involvement.

Context: Reddit licenses content to companies like Google and OpenAI, and filed a similar case against Anthropic in June.