Skip to main content

The news desk

AI news, checked against the source.

What changed in AI, one story at a time. Each one links to its primary source, and facts stay apart from interpretation.

Last updated

Monday, October 5UTC

  1. New
    RekaRho-1Video

    Reka previews Rho-1 for conversational image and video editing

    Reka announced a research preview of Rho-1, a 19B model that it says can understand and generate text, images and video within one conversation. Its demo shows image creation, video animation and a weather edit across turns. The original page does not establish broad app or API access or independent output quality.

    Reka1 source

Free, email only

Unlock the full brief free.

  • Thirty days of stories to browse, not seven
  • On every story: what to check, each source and why it matters
  • The same signup unlocks every free tool

This is not an account: there is no password, and the unlock is a cookie in this browser. You also join the Rise Productive newsletter from Demetri Panici, about once a week: what I built and what changed in AI. We'll email you a link to confirm, and you can unsubscribe in one click. The same signup unlocks every free tool on the site. How your email is handled.

Already subscribed? Enter the same email to unlock this browser. You won't be signed up twice.

Sunday, October 4UTC

  1. New
    HeyGenHyperframes 0.8.120Video

    Hyperframes improves CLI publishing login recovery

    Hyperframes 0.8.120 handles refused or expired publishing logins by refreshing access or asking the user to sign in instead of leaving a broken publishing state. It also stabilizes cross-file element IDs and separates feedback labels for people and agents. This is a tool reliability release, not a new video model; opening the news story does not publish anything.

    HeyGen1 source

  2. New
    HeyGenHyperframes 0.8.119Video

    Hyperframes fixes freeze-frame undo and duplicate preview actions

    Hyperframes 0.8.119 fixes removal of a freeze-frame still when extraction finishes after an undo, and avoids repeating playhead actions during preview redraws. This is a version-specific editing reliability update, not a new generation model or guarantee that every editing bug is resolved.

    HeyGen1 source

  3. New
    HeyGenHyperframes 0.8.118Video

    Hyperframes adds ad-template previews and aligns animation playback

    Hyperframes 0.8.118 adds video previews for eight ad templates and aligns GSAP and MotionPath playback with the composition preview. It also builds host controls before rendering thumbnails. These are template and editing improvements, not a new AI video-generation model.

    HeyGen1 source

  4. New
    HeyGenHyperframes 0.8.117Video

    Hyperframes fixes lost edits and fails visibly on broken video frames

    Hyperframes 0.8.117 repairs persistence and redraw behavior for resize, rotation and undo changes to animated layers. It also fails visibly when a video frame cannot be decoded rather than silently exporting blank footage. This is a concrete editing reliability update, not a new video-generation model.

    HeyGen1 source

Saturday, October 3UTC

  1. New
    HeyGenHyperFrames 0.8.115Video

    HyperFrames exposes harness usage to agents and steadies video preview audio

    The release adds a CLI usage command for Claude Code, Codex and Grok harness subscriptions. It also keeps music playing across preview cuts, forwards Kokoro TTS speed and closes owned browser processes when the CLI stops. This is video-tooling and harness integration work, not a new generation model or larger provider allowance.

    HeyGen2 sources

  2. New
    Builder.ioClips nightly 0.1.423-0Video

    Clips nightly isolates malformed Lab preferences instead of blocking unrelated settings

    The published nightly includes per-setting error handling and repair controls for invalid Lab choices. The change is a reliability repair in creator tooling, not a new video-generation model or the earlier recording-startup fix.

    Builder.io2 sources

  3. New
    Builder.ioClips nightly 0.1.422-0Video

    Clips nightly improves recording startup, countdowns and cancelled-attempt cleanup

    This nightly restores concurrent recording setup and starts the countdown alongside preparation. Cancellation cleans up that attempt's pending Rewind setup and audio leases. Official tests cover the repaired startup paths, including older macOS behavior; this is capture tooling, not a video-generation model.

    Builder.io2 sources

  4. New
    HeyGenHyperFrames 0.8.114Video

    HyperFrames isolates file renames and keyframe edits instead of changing unrelated animation

    Version 0.8.114 fixes overly broad file-reference rewrites and keyframe edits that copied other tweens' channels. It also handles modern CSS colors and improves frame capture and preview-server ownership.

    HeyGen1 source

Friday, October 2UTC

  1. New
    HeyGenHyperFrames 0.8.113Video

    HyperFrames repairs export-layer blinking and frame-zero animation

    The release repairs animation set steps landing on their intended frames, including frame-zero intros. Its CLI rejects stray positional arguments and preserves an explicit port for background previews. These are rendering and workflow fixes, not generative-model quality claims.

    HeyGen GitHub1 source

  2. New
    HeyGenHyperFrames 0.8.112Video

    HyperFrames previews new timeline tracks and decouples saves from thumbnails

    Studio previews where a dragged clip will create a new track and no longer waits for thumbnails to save. The release also adds timeline inspection details and a motion-blur skill, alongside CLI and render checks.

    HeyGen GitHub1 source

  3. New
    CreatifyBoreal-H3Video

    Creatify launches Boreal-H3 for product-consistent ad videos

    Creatify announces an advertising-focused model post-trained from MiniMax H3, with text, image, and reference-to-video modes in its playground and API. Published list prices are $0.04 per video-second at 768p, $0.12 at 1088p, and $0.26 at 2K. Supplier evaluations emphasize product and creator consistency; they do not prove improved real campaign conversions.

    Creatify AI via PRWeb1 source

  4. New
    HeyGenHyperFrames 0.8.111Video

    HyperFrames makes host undo immediate during timeline saves

    Embedded Studio hosts can repaint a still-saving clip move immediately when their own undo key is pressed. The release is a focused editing-recovery change, not a new video model.

    HeyGen GitHub1 source

  5. New
    Builder.ioClips nightly 0.1.417Video

    Clips nightly keeps recording controls visible when you move the camera

    The new nightly includes a focused desktop fix: repositioning the camera bubble no longer schedules dismissal of the recording controls. This addresses the pre-recording controls disappearing, not the separate recording-storage recovery covered in nightly 416.

    Builder.io2 sources

  6. New
    Builder.ioClips nightly 416Video

    Clips nightly lets creators record before connecting storage and recover interrupted uploads

    Clips nightly 416 contains the record-first change: creators can capture before linking Builder or S3 storage, with a local copy retained until a confirmed upload. The implementation adds browser reload recovery, desktop pending-upload handling, deletion confirmation and account-ownership checks. Failed connections and offline uploads keep the local copy. This is a creator recording workflow, not a new generative-video model or an unconditional never-lose guarantee; desktop and extension runtime behavior were covered by maintainer unit tests, not live runtime trials.

    Builder.io GitHub2 sources

  7. New
    RunwaySeedance 2.5 Draft Mode in AgentVideo

    Runway Agent adds fast Seedance drafts before enhancement

    All plans can use 480p Draft mode in Agent, then enhance the result. The toggle is automatically suggested for faster iteration.

    Runway1 source

Thursday, October 1UTC

  1. New
    MixVioMixVio AI workspaceVideo

    MixVio launches a connected workspace for image, video and audio production

    MixVio announced a creative workspace that lets a generated image continue into editing or image-to-video without repeated downloads and uploads. It brings model-compatible image, video and audio workflows under one account and credit system, rather than introducing a new foundation model. Its publisher says quotes appear before generation and failed tasks release reserved credits. Input and output limits vary by selected route; account behavior was not tested.

    MixVio via PR Newswire2 sources

  2. New
    RunwayRunway for ChatGPT DotsVideo

    ChatGPT dots can use the connected Runway plugin to make videos

    Runway documents a ChatGPT plugin that a dot can also use after setup. The agent plans the shots and delegates media generation to Runway; no separate dot-specific connection is required.

    Runway2 sources

  3. New
    TavusGriffin-LiteVideo

    Tavus previews a model that listens, sees and responds while speaking

    Tavus introduced Griffin-Lite, a full-duplex video-to-video research preview. Its concurrent perception and speech-and-video generation are designed to react to interruptions, expressions and visual context during a conversation. Access is restricted to selected trusted testers; Griffin is not yet on the Tavus platform or available to customers. The publisher is developing disclosure and safety measures before broader release.

    Tavus1 source

Wednesday, September 30UTC

  1. New
    RunwayRunway AdsVideo

    Runway Ads connects creative generation, approval, publishing and performance feedback

    Runway announced an agentic advertising product that generates branded image and video variants, publishes approved ads to Meta, Google and TikTok, and uses platform results to generate the next round. It adds localization, format changes, brand checks and budget guardrails. Access is an enterprise pilot with an early-access request, not a broadly available self-serve launch.

    Runway2 sources

  2. New
    Utopai StudiosPAI and Utopai XVideo

    Utopai connects production agents, shot history and its video model in PAI

    Utopai Studios announced an updated PAI workspace and its Utopai X video model. Production Assistants use scripts, character and location assets to help organize scenes, plan shots and generate iterations. Takes and references remain connected to their shot for review and revision. Filmmakers choose models, control generation costs and make final decisions. Utopai X builds on an existing video foundation model; the platform is not an autonomous finished-film promise.

    Utopai Studios2 sources

Days are UTC. Some sources report an exact time and others only a date; both are placed on their UTC day.