Claude Down: Consumer users disrupted while Anthropic says business API stayed live

Why this matters now: The claude down incident interrupted consumer access across web and apps at scale while the company’s business-facing API continued running, creating a split impact between individual users and enterprises. Anthropic later indicated services were restored, but the outage and recent demand surge underline pressure on core systems.

Who felt the impact first and how

Consumer-facing users were the most immediately affected: the outage made Claude unavailable on public surfaces such as claude. ai and mobile apps, and users experienced failed requests, timeouts or inconsistent responses. At the same time, the Claude API that powers business customers remained operational, leaving enterprises largely insulated from the consumer disruption.

Claude Down: Status timeline and signals

The incident was flagged on March 2, 2026. An initial investigating notice appeared at 11: 49 a. m. GMT (6: 49 a. m. ET), mirrored elsewhere as 11: 49 UTC, with a follow-up update at 12: 06 UTC noting the investigation was ongoing. Later in the day, an update at Mar. 2, 2026, 4: 37 p. m. UTC indicated that Claude was back up across claude. ai and the company’s apps while teams continued to match unusually high demand.

• 11: 49 a. m. GMT (6: 49 a. m. ET / 11: 49 UTC): investigation posted.
• 12: 06 UTC: follow-up update confirming the investigation continued.
• Mar. 2, 2026, 4: 37 p. m. UTC: company update that consumer services were restored.

Systems affected and scope

Errors were described as "elevated errors" and affected multiple consumer-facing products: the core Claude chatbot, the model identified as Claude Opus 4. 6 (the latest-generation LLM), the developer-facing console referred to as Claude Console, and the coding assistant labeled Claude Code. The company also noted problems tied to claude. ai login/logout paths and that some API methods were not functioning for consumer surfaces, while the Claude API used by businesses remained unaffected.

Operational reports from a monitoring service recorded over 2, 000 user reports of problems during the outage window. The material also notes a prior episode of elevated errors on Saturday, Feb. 28, 2026, which affected Claude Opus 4. 6 and was found and resolved.

Marketplace context and an unsettled political note

The outage arrived shortly after Claude reached number one as the most-downloaded free app in the U. S. app store, displacing a close competitor that ranked second while another competitor sat fourth. Free Claude users had increased by more than 60 percent since January—a surge the company described as unprecedented demand in its recovery messaging.

The available material also references a recent political sequence: an announcement that the U. S. government would no longer use Anthropic technology, followed by a deal for a rival AI provider to supply technology to the U. S. Department of War for military use in "classified environments. " This element appears in limited coverage and should be treated as developing while it is verified further.

Recovery status and what could confirm stability

Anthropic’s later messaging indicated consumer surfaces were restored and teams remained focused on matching demand. For users and developers, confirmation of a return to normal will be a continued absence of elevated error reports, stable login/logout behavior, and full functionality of previously affected API methods. The material available did not include an ETA while the incident was active.

• Key indicators to watch for: sustained low error rates across consumer surfaces, restored login/logout flows, and no new model-specific incidents.
• If problems reappear, expect follow-up investigation notices and targeted fixes for API methods tied to consumer features.

Here’s the part that matters for users: consumer services were the primary casualty, while enterprise integrations remained online, which changes how customers should prioritize mitigation steps.

• Service restored later the same day, per the company update at Mar. 2, 2026, 4: 37 p. m. UTC.
• Consumer surfaces—claude. ai and mobile apps—were unavailable during the outage window.
• Claude Opus 4. 6, Claude Console, and Claude Code were listed among affected products.
• A monitoring site recorded over 2, 000 user problem reports during the incident.
• Previous elevated errors occurred on Feb. 28, 2026, affecting Claude Opus 4. 6 and were resolved then.

It’s easy to overlook, but the simultaneous surge in downloads and user growth adds pressure to reliability work; repeated stress events can reveal deeper capacity or routing issues that need structural fixes.

Writer’s aside: rapid demand spikes make consumer and developer surfaces behave very differently in incident windows, and restoring login paths is often the trickiest part of recovery even when backend compute remains healthy.

If you’re wondering why this keeps coming up: the company framed the outage as tied to unprecedented demand and specific consumer-path failures while keeping business APIs live, which helps explain the split experience between individual users and enterprise customers.