Claude outage triggers “Unable to authenticate” errors — resolved after one-hour disruption

Claude users hit “unable to authenticate” errors; Anthropic says outage is resolved

A brief but widespread disruption hit Anthropic’s Claude services on Tuesday, September 29, 2026, leaving many users facing log-in failures and the on-screen warning “Temporarily unable to authenticate. Please retry.” Anthropic’s status page shows the incident affected claude.ai (web, desktop, and mobile), Claude Code, Claude Cowork, and the Claude API between 07:00–07:59 PT (14:00–14:59 UTC) before service recovered. The company reports systems have been operating normally since recovery. (anthropic.statuspage.io )

What happened and when

Anthropic posted a rolling incident timeline Tuesday morning noting “elevated error rates” across multiple services. After initial mitigation at 14:36 UTC, a second issue prevented many users from signing in—specifically impacting single sign-on (SSO) and Sign in with Apple—before the broader fix was applied. The incident was marked resolved at 16:27 UTC with service restoration pegged to 14:59 UTC (07:59 PT). Users may find that some messages sent during the 14:00–14:59 UTC window were not saved, according to the update. (anthropic.statuspage.io )

What users saw

While the outage was unfolding, users reported repeated authentication failures and session drops across the web app and desktop, with many seeing the banner “Temporarily unable to authenticate. Please retry.” Others noted that existing sessions sometimes continued to run while new chats or purchases failed—consistent with Anthropic’s advisory to avoid signing out during the incident. (reddit.com )

Scope and impact

The disruption touched both consumer and developer surfaces: claude.ai, the Claude desktop and mobile apps, Claude Code and Cowork, plus the API and Console. Third‑party monitors and news outlets tracked a sharp spike in reports during the 07:00–08:00 PT hour, with coverage noting thousands of user complaints and confirming Anthropic’s acknowledgement on its status dashboard. (anthropic.statuspage.io )

Service status now

As of publication, Anthropic’s status dashboard lists “All Systems Operational,” and the September 29 incident appears in the Past Incidents log with a final “Resolved” note. If you’re still seeing sign‑in failures or stuck sessions, Anthropic advises standard troubleshooting (below). (anthropic.statuspage.io )

Troubleshooting if you still can’t authenticate

If residual issues persist on your device or account after a platform outage clears, these steps may help:

• Check official status first to confirm there isn’t an active incident before changing local settings. (anthropic.statuspage.io )
• If you’re already signed in and service is degraded, avoid signing out until the status page confirms recovery; this mirrored Anthropic’s guidance during the incident. (anthropic.statuspage.io )
• For email “magic link” sign‑ins, open the link on the same device/browser that requested it; otherwise you may be prompted for a verification code and fail to complete the flow. (support.claude.com )
• Clear browser cache/cookies or try an incognito window, then retry the sign‑in; Anthropic’s help center lists these among standard remedies for generic login errors. (support.claude.com )
• If your organization uses SSO and it’s misconfigured or temporarily unavailable, contact your IT admin or try an alternate approved method. Anthropic documents SSO-specific requirements and known failure modes. (support.claude.com )
• Developers: verify your API key authentication from the Console, and confirm your requests aren’t failing on expired credentials or rate limits after service restoration. If platform-wide issues appear again, pause re‑auth attempts and monitor the status feed. (anthropic.statuspage.io )

How this compares to recent hiccups

September has seen a handful of smaller incidents across Claude services. On September 22, Anthropic reported elevated errors across several model families for about 80 minutes. A week earlier, on September 15, the company logged intermittent spikes affecting Claude Mythos 5.1 and Fable 5.1. Neither of those prior events centered on widespread authentication failures, but together they show a month with intermittent reliability issues. (anthropic.statuspage.io )

Reporting from the field

User communities surfaced the auth failures minutes before the status page reflected the scope, posting screenshots and short clips of failed log-ins and repeated banner warnings. Some posts noted the paradox of a “green” status display while DownDetector-style graphs spiked; the official page was updated shortly thereafter with the broader incident classification and timeline. Such gaps are common early in fast-moving outages, as providers validate telemetry and scope before labeling an incident. (reddit.com )

What we still don’t know

Anthropic has not attributed a technical root cause for Tuesday’s disruption beyond describing elevated error rates and a separate sign‑in failure affecting SSO and Sign in with Apple. There is no indication on the status page of a security incident, data loss beyond unsaved messages during the 14:00–14:59 UTC interval, or user action required beyond standard troubleshooting. We’ll update this story if the company publishes a post‑mortem. (anthropic.statuspage.io )

Bottom line

• Timeline: 07:00–07:59 PT (14:00–14:59 UTC) on September 29, 2026. (anthropic.statuspage.io )
• Symptom: “Temporarily unable to authenticate” and failed sign‑ins; SSO and Sign in with Apple were briefly unavailable. (reddit.com )
• Status: Resolved; Anthropic reports all services operating normally. Some messages during the outage window may not have saved. (anthropic.statuspage.io )

If you’re still affected, confirm the status page shows no active incident, then follow the help‑center sign‑in guidance and SSO checks above. (anthropic.statuspage.io )