AI news: ElevenLabs SFX v3, ChatGPT account switching and Claude Code updates — 31 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

ElevenLabs has introduced SFX v3 for sound-effects workflows, ChatGPT has begun rolling out mobile account switching, and Claude Code 2.1.316 documents the approaching retirement of Opus 4.8 fast mode. Kimi adds session import, ServiceNow distinguishes a live plugin from future integrations, and OpenAI announces support for a forthcoming shared AI-training preference signal.

This edition covers 28 October at 12:06 through 31 October at 11:55 Europe/Istanbul (UTC+3). Announcement dates below are source dates, not assumed release times. The separate incident edition covers service disruptions and recoveries.

ElevenLabs SFX v3: multiple sounds, music and vocals in one effects prompt

ElevenLabs announced SFX v3 on 29 October and updated the announcement on 30 October. It says the sound-effects model is available in ElevenCreative and ElevenAPI, with creation through the Sound Effects tab, Flows, Studio and Music Editor, plus the mobile app. Original SFX v3 announcement.

The vendor describes more reliable handling of prompts containing several sounds, stronger music and vocal elements, and fewer unwanted artifacts. Those are ElevenLabs’ quality and benchmark claims, not MUVO test results. Creators can describe a combined sound scene, including musical or vocal elements; this is not a replacement launch for ElevenLabs Music.

The announcement does not establish a new price, universal plan entitlement or a complete regional availability list. Check the relevant account or API configuration before assuming access or costs.

ChatGPT: mobile account switching begins rolling out

The 30 October release notes announce account switching on iOS and Android. Users can keep up to two accounts signed in and switch the active account through Settings. This is a rollout, so the announcement does not establish simultaneous availability on every device. ChatGPT release notes.

OpenAI’s help article says each account keeps its own chats, memory, files, settings, workspaces and subscriptions. Switching does not merge accounts or pool their subscription benefits. For someone separating personal and work activity, the practical benefit is less repeated signing in while retaining separate account contexts. Organizational restrictions can still apply. Account-switching guide.

Claude Code 2.1.316: fast-mode retirement and workflow fixes

Claude Code 2.1.316 was published on 31 October at 04:01 Istanbul / 01:01 UTC. Its release notes say Opus 4.8 fast mode was deprecated on 30 October and will be removed on 30 November. After removal, those fast-mode requests will error rather than fall back to standard mode; Anthropic directs users needing fast mode to Opus 5.5. Official release.

The release also documents fixes for restricted files outside the working directory, hangs after truncated compressed network responses, and VS Code sessions incorrectly returning to Auto mode after the user chose manual edit approval. Cloud sessions gain visible forms for MCP requests that previously could be cancelled without being shown.

Practical impact: teams using the deprecated fast-mode combination have a dated migration dependency. The listed fixes are vendor release notes, not proof that MUVO reproduced each defect or that every hosted surface updates at the same instant.

Kimi Code: import existing coding sessions and control planning explicitly

Kimi’s 28 October notes introduce CLI 2.4.0 and Desktop 1.2.0. The CLI can import recent Claude Code and Codex sessions from a 15- or 30-day window, then continue them as Kimi sessions. Plan mode now starts only when enabled by the user, and plans require review even under Never Ask. Tower gains model selection per task. Official Kimi Code notes.

Desktop 1.2.0 adds an optional sleep-prevention setting and automatic mentions for pasted images. These are client workflow changes; the announcement does not establish a new model or a change to subscription prices. Analysis: session import can reduce migration work, but imported history should not be assumed to reproduce another tool’s exact execution state.

ServiceNow and OpenAI: a plugin now, broader integrations later

The 28 October partnership announcement says the ServiceNow MCP plugin is available in OpenAI’s marketplace. Customers connect their own instance and select exposed capabilities; actions follow the organization’s ServiceNow rules and permissions. Original announcement and availability schedule.

The same announcement places enhanced plugin capabilities by the end of 2026, voice features in the first quarter of 2027, security early access in January 2027 with general availability planned for the second quarter, and Codex in AI Workflow Factory in 2027. These are planned dates, not confirmation those capabilities are available today. Interactive ChatGPT experiences are described as previews.

For enterprise users, the useful distinction is between the existing plugin and future voice, security and development integrations. The announcement does not establish consumer access to a ServiceNow instance or automatic permission to change its records.

AI training preferences: OpenAI supports a shared signal, implementation still ahead

On 30 October, OpenAI said it will honor a shared voluntary robots.txt signal for foundational generative-AI training preferences. It aims to implement support in the coming months; existing crawler controls continue to apply. OpenAI announcement.

The linked proposal is an individual Internet-Draft, updated 30 October, with no formal standing in the IETF standards process. It distinguishes content-use preferences from crawler traffic rules. It is not a finalized internet standard. Original draft and status.

Analysis: a common signal could reduce the number of provider-specific preferences publishers maintain, but its practical effect depends on implementation and participation. The announcement does not establish retroactive removal of training data or replace creators’ existing licensing decisions.

In a separate 30 October advocacy post, OpenAI highlighted briefs supporting its view of fair use in AI training in the New York Times litigation. That is an attributed company position and selection of submissions, not a new court ruling or a licensing permission for creators. OpenAI’s post.

Google Project Suncatcher: an orbital AI research milestone

Google reported on 30 October that its prototype satellite returned an AI response from a data-center-class TPU running Gemma 4. It describes commissioning and early machine-learning tests after the 1 October launch, with thermal/radiation testing next and twin interconnected satellites targeted for 2027. Google’s original report.

This is a vendor-reported research milestone for AI computing infrastructure. It is not a new Gemini subscription feature, production API region or announced reduction in users’ generation prices.

Sources

• ElevenLabs: SFX v3.
• OpenAI: mobile release notes, ServiceNow schedule and training preferences.
• Anthropic and Kimi: Claude Code 2.1.316 and Kimi Code releases.
• Google: Project Suncatcher update.

We checked the wider creative-tool and major-provider watchlist, deduplicating previously covered announcements. The 30 October edition retains Pika H3 Flex, Suno Studio and Mureka MP4 coverage. Undated marketing copy, isolated complaints and speculative model-release posts were not treated as launches. Performance, availability and future schedules above are attributed to original sources; MUVO did not conduct paid generation tests.

MUVO publishes this reporting and offers competing creative tools. See our reporting method.

Update timeline

• 31 October 2026, 11:55 Istanbul / 08:55 UTC: Research cutoff. Added verified releases, rollouts and dated announcements from 28–31 October, keeping available products, future integrations, research and attributed policy debate distinct.