AI news: Suno Studio, ElevenLabs Brand Kit and Google Gemini agent updates — 9 October 2026

Publisher disclosure: MUVO offers competing music, remix, and video tools. This report summarizes public sources and provider documentation; it is not an independent product test or a claim that MUVO is free of similar issues. Read our reporting method.

Key product updates at a glance

Today’s creative-tool changes focus on editing control and reusable visual identity; the major-provider announcements focus on agents and developer workflows. This edition covers sources checked between 6 October at 12:17 and 9 October at 11:58 Europe/Istanbul (UTC+3), using overlap to catch missed announcements. The edition date is not the launch date of every item.

Service failures and recoveries are in the separate 9 October incident report.

Suno Studio: more control over timing, stretching and fades

Suno’s 8 October update adds editing controls to its browser-based Studio, included with Premier. Shift-click-drag now stretches a clip in time, while a draggable fade control changes the curve beyond a linear fade. The wavetable synth, musical typing keyboard and expanded plug-in views can open in floating windows. Original Studio release note.

Suno also says it improved first-downbeat detection, including for material such as a cappella vocals, to reduce manual alignment. That is the vendor’s claim; we have not measured it. For producers arranging samples, the practical change is finer control over how clips meet on the timeline. The announcement does not establish a new generation-model version or resolve the separate customer quality concerns in our Suno report.

ElevenLabs Brand Kit: visual identity becomes reusable workspace context

ElevenLabs says Brand Kit is available in ElevenCreative, with colors, fonts, logos, reference images and rules stored together and referenced using @ in Image & Video, Flows and the ElevenCreative agent. Teams can share a workspace kit or maintain different kits for different brands. Original Brand Kit announcement.

This is a workflow change for repeated image and video creation; consistent output remains something to evaluate, not a result MUVO has tested. The announcement does not specify a complete plan or regional eligibility list. Voice and sound extensions are future plans, not part of the visual feature described as available.

The public page was accessible at our cutoff and displays 9 October as its publication date, alongside an 8 October update date. Its machine-readable publication time is later than our retrieval. We therefore use the visible date and observed availability statement without asserting an exact launch time.

ElevenReader: Brazil launch separates free access from the Ultra catalog

ElevenLabs announced ElevenReader’s Brazil launch on 8 October, including Fábio Porchat’s voice and Brazilian Portuguese books supplied through Bookwire Brasil. The reader app is free on iOS and Android; the announcement places the full book catalog behind ElevenReader Ultra. Brazil launch announcement.

A voiced edition of Dom Casmurro is offered free to all users. Reading imported PDFs, articles and ebooks is a consumer listening workflow; this announcement does not grant a general commercial license to generate with the named performer’s voice. Check the app’s subscription and title availability for the material you want to hear.

Google Gemini agent: an enterprise announcement with industry previews

Google announced Gemini agent at Gemini at Work on 8 October, describing one agent and API for knowledge work, content and media creation, and coding. It connects to business systems, runs longer tasks and uses enterprise controls. These are Google’s descriptions, not MUVO performance or security findings. Google Cloud announcement.

Google names Gemini and Anthropic Claude as model choices, with additional providers planned. Financial Services and Legal specializations are in preview; Government, Healthcare and Retail are coming later. The announcement does not provide a complete plan, country or account rollout schedule for the overall agent.

For creative teams, the stated appeal is connecting briefs, assets and execution. Confirm organizational access before planning a migration. This source does not establish a new consumer Veo, Flow, Lyria or MusicFX release; the previous edition’s personal Gemini changes concern a different product surface.

OpenAI Codex: faster steering rolls out in the desktop app

OpenAI’s 8 October release note announces faster steering for Codex in the ChatGPT desktop app. Follow-up messages can redirect a running task sooner. Settings → General → Follow-up behavior controls whether they steer the current run or wait for the next one. Original release note.

This can help when a developer corrects an approach or adds missing context mid-task. OpenAI describes a rollout, without a complete plan or region matrix in this entry. It is a workflow update, with no new model established by this notice.

Anthropic: November policy changes and a new security program

Usage Policy: announced now, effective 12 November

Anthropic published its policy update on 8 October with an effective date of 12 November. It consolidates restrictions on deceptive campaigns and artificial activity, clarifies high-risk human-review requirements, and adds controls for autonomous physical actions. Original policy announcement.

Its new restriction on sustained, needless abuse of models explicitly excludes ordinary frustration, pushback, dark creative themes and model testing or research. That qualification matters for writers and developers evaluating what changes. Teams should review the complete policy for their use case; the announcement does not mean every new provision took effect on 8 October.

Cyber Mission: free scans for eligible open-source maintainers

Anthropic’s 8 October Cyber Mission announcement introduces OSS Scanner and a Critical Infrastructure Defense Program. OSS Scanner offers free, periodic scans to core maintainers of critical open-source projects who can triage findings. Its reports are model-generated without human review and can contain inaccuracies. Cyber Mission announcement.

The infrastructure program starts with a small provider cohort and offers an interest-registration route. These are targeted security programs, not a general entitlement for every Claude subscriber. For teams maintaining creative software, the useful next step is to check enrollment criteria and plan human review of any findings.

Other verified developer updates

Microsoft Copilot Studio: September changes summarized on 7 October

Microsoft’s 7 October post is a roundup of September updates, not evidence that all features launched this week. It puts Apps builder and Managed Runtime in public preview, while Foundry IQ integration and the Review panel are generally available. Copilot Studio support for the centralized plugin registry is rolling out over the coming weeks. Original Copilot Studio roundup.

For teams extending business agents, the release stage should shape adoption plans. A preview listed beside an available integration does not inherit its availability. The post does not establish unrestricted access for every Microsoft account.

Perplexity: retrieval and decision-model releases

Perplexity’s official 7 October forum announcement releases two pplx-embed-v2-late models for retrieval across text, images and pages, with a shared embedding space across model sizes. The announcement says they are publicly available on Hugging Face. We have not independently verified its performance claims. Original embedding announcement.

A separate 6 October post announces pplx-decider-v1.1-27b, an updated open-weights multimodal decision model. These are developer releases, not new consumer image or video generators. Review the model’s own documentation and license before incorporating it into a product. Original decision-model announcement.

Analysis: the workflow change may matter more than a new model

MUVO’s interpretation: today’s clearest creator-facing gains are more direct editing control and reusable context. Enterprise announcements extend that theme to connected systems, but often leave account eligibility or rollout timing unresolved. Match a feature to a real workflow and confirm access before making it a dependency. A vendor announcement establishes what was announced; it does not substitute for testing with your own material.

Sources

• Suno: Studio improvements, 8 October.
• ElevenLabs: Brand Kit and ElevenReader Brazil.
• Google: Gemini at Work announcement.
• OpenAI: 8 October ChatGPT release notes.
• Anthropic: Usage Policy announcement and Cyber Mission.
• Microsoft: Copilot Studio September roundup.
• Perplexity: embedding models and decision model.

We also searched the remaining music, video and remix/stem watchlist, including Udio, Mureka, SOUNDRAW, AIVA, Beatoven.ai, Runway, Kling, Pika, Luma, Moises, Fadr and Kits AI, plus Adobe, ByteDance and Stability AI. No additional fresh release qualified in the accessible evidence. Search coverage is selective, and unavailable sources are recorded in the research ledger. See our reporting method.

Update timeline

• 9 October 2026, 11:58 Istanbul / 08:58 UTC: Initial edition consolidates ten original announcements or release entries, including newly discovered items from the overlap window. Announcement, preview and future-effective dates are kept distinct. No paid generations or private account-level product tests were performed.