AI incidents: ElevenLabs Music, Claude reporting and OpenAI recoveries — 8 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.

Incident summary and status at publication

Update, 14 October: all listed events now carry provider recovery notices. The original edition recorded Claude reporting under monitoring. These are provider-confirmed events with different scopes, not evidence of one industry-wide outage.

This edition covers 5 October at 11:58 through 8 October at 12:17 Europe/Istanbul (UTC+3), using a 48-hour overlap with the last successful scan. Unchanged events from earlier editions are deduplicated. Every event time in this article is Istanbul time. Times below describe posted updates unless explicitly labeled otherwise; they do not establish actual outage durations. Releases are covered in the separate 8 October news edition.

ElevenLabs Music 2.5: recovered after an infrastructure failure

ElevenLabs says its Music service recovered on 8 October. Its initial 11:48 notice identified degraded Music endpoints while saying Music v2 should work. The title specifies Music 2.5; the component is listed under ElevenAPI. At 11:51, it reported a fix and recovery monitoring. At 11:58, it marked the incident resolved. Original Music incident.

The provider attributes the failure to a dormant infrastructure configuration problem triggered by routine maintenance. New instances could not start, automated recovery failed, and staff restored service manually. Actual impact boundaries are unspecified; chart and post times are not used to calculate duration. The fallback statement promises neither job recovery nor refunds.

Claude: usage reporting and spend-limit pauses resolved

Usage data: recovery confirmed on 13 October

Anthropic says API requests are unaffected by this usage-reporting incident. It acknowledged errors on the platform.claude.com Usage page and Admin API usage report at 7 October, 16:25. An update at 19:36 also named the Rate limits page. At 20:28, Anthropic said the Admin API report had recovered and fixes for the console pages were progressing. Usage-data incident.

At 8 October, 00:44, it moved to monitoring: errors were mitigated, but roughly 30 minutes of recent usage could be delayed. A full fix remained in progress. This is a separate notice from the earlier usage-report incident covered yesterday. No common technical cause or exact impact start has been published. Delayed reporting should not be read as evidence that no API usage occurred.

Update, 14 October: Anthropic resolved this reporting incident on 13 October at 21:00 Istanbul. See the recovery roundup.

Incorrect spend limits: some organizations had requests refused

Anthropic says some organizations were incorrectly paused for reaching a spend limit, causing refused requests across the Claude API, claude.ai, Claude Code, Claude Security and Claude Cowork. Its only visible update reports resolution at 8 October, 00:23. Actual start, recovery time and duration are not specified. The provider describes an erroneous limit check; it does not say every organization was affected. Spend-limit notice.

OpenAI: desktop recovery and four separate service incidents

ChatGPT Work desktop: install the fixed version

OpenAI now directs affected users to update their desktop app. Yesterday’s edition tracked version 26.1002.51308, which could not create new Work threads. On 7 October at 12:28, OpenAI announced a fix for Linux and macOS; at 14:53, it confirmed a fixed release for Windows as well. The notice does not name the replacement version. Desktop incident and update advice.

We have updated the 7 October article in place. This is recovery of the previously reported issue, not a new desktop outage today.

Dot workflows: two distinct notices

• Creating threads through Dot in Codex: investigation was posted 7 October at 19:31, mitigation/monitoring at 20:30, and resolution at 21:06. Thread-creation notice.
• Dot turn failures, delayed replies and reduced proactivity in Codex and Work: investigation was posted 7 October at 20:21 and resolution at 21:23. Performance notice.

Both list the dots component. Their timelines overlap, but OpenAI has not provided a common cause. Neither establishes that every Codex or ChatGPT workflow failed.

FedRAMP workspaces and APAC users: recovered

GPT-5.6 Instant in FedRAMP workspaces: OpenAI acknowledged elevated errors at 8 October, 02:42, reported mitigation and no further observed errors at 02:48, and posted resolution at 03:03. This is a model-and-workspace-specific event. FedRAMP notice.

Some APAC users: a separate notice concerned ChatGPT Work, conversations and GPTs. Investigation was posted at 8 October, 06:17 and recovery at 07:51. No country list, affected-user percentage or technical cause is supplied. APAC notice.

These updates do not provide sufficiently explicit actual impact boundaries to calculate outage durations.

What affected users can check

Use the product, region, model, app version and source timeline to decide whether a notice matches your problem. For Work desktop, follow OpenAI’s instruction to install the latest version. Claude reporting is now marked resolved; consult the linked notice if a discrepancy persists.

MUVO’s practical suggestion: inspect an existing task or generation before starting another paid attempt, and retain its request ID and error text for support. A provider marking an incident resolved does not prove that a specific interrupted job or account recovered. None of the notices above promises automatic refunds.

Customer-review follow-up and coverage limits

The scan also checked music, video and remix/stem reviews and rechecked the existing Suno vocals, Runway retries and credits, Moises lyrics and Kits download-minutes reports. No verified resolution or material change to those reports was established. Repeated quality complaints and isolated new access or credit allegations remain research leads. Matching Suno posts by the same reviewer across two listings were counted once.

Google coverage is incomplete. Google Cloud’s history yielded no qualifying new AI event in the reviewed material, while the separate Gemini API status page remained inaccessible. Cloud history cannot establish consumer Gemini, Flow, Veo or MusicFX availability. Other gaps include inaccessible review permalinks, AIVA/Beatoven review pages, and history interfaces that exposed no readable events for some providers. This daily scan is not continuous monitoring or a complete incident census.

Update timeline

• 8 October 2026, 12:17 Istanbul / 09:17 UTC: Initial edition includes seven new official notices and the desktop recovery follow-up. Yesterday’s desktop coverage was updated; Claude usage reporting remained under monitoring at the original cutoff.

Sources

• ElevenLabs: Music 2.5 incident.
• Anthropic: usage-data incident and spend-limit issue.
• OpenAI: desktop fix, Dot thread creation, Dot performance, FedRAMP and APAC.
• Review and policy evidence is linked in the four existing customer reports above. Reporting method explains the different standards for official notices and customer allegations.

Original notices were checked on 8 October 2026. Provider statements establish what the operators acknowledge; MUVO did not independently measure these services or test paid generations.

• 14 October 2026, 13:26 Istanbul / 10:26 UTC: Added the later Anthropic reporting recovery; original publication date and URL preserved.