ChatGPT, Claude, Gemini down: What we know about the AI outages (updated)
A sudden major AI outage disrupted millions of digital workflows on Thursday, temporarily knocking out major artificial intelligence services around the world. Users attempting to generate text, write code, or execute queries found themselves facing unexpected error messages and system downtime across multiple leading platforms.
While brief service hiccups occasionally happen for individual artificial intelligence tools, a simultaneous disruption affecting virtually every market leader is exceptionally rare. In this comprehensive breakdown, we examine what caused the major AI outage, which AI service providers were hit, how the recovery unfolded, and what companies are doing to prevent future infrastructure failures.
What Happened During the Major AI Outage?
On Thursday morning, digital monitoring services and social media platforms began lighting up with reports of widespread connectivity issues. Thousands of users discovered that their primary artificial intelligence tools were either completely unresponsive or severely degraded in performance.
According to tracking data from DownDetector, user error reports began spiking dramatically around 11:00 a.m. Eastern Time. The surge in complaints signaled that the issue was not isolated to a single geographic area or internet service provider.
Instead, the disruption impacted multiple competing service platforms at the exact same moment. For several hours, professionals, developers, students, and casual users struggled to access their preferred tools for daily tasks.
As news of the incident spread across social media, speculation mounted about whether the problem stemmed from shared cloud infrastructure, server facility disruptions, or unexpected network routing glitches.
Which AI Service Providers Were Affected by the Downtime?
The extent of the disruption was particularly striking because it crossed competitive lines, hitting nearly all major enterprise and consumer platforms at once. Among the platforms experiencing stability and downtime issues were:
- ChatGPT (OpenAI): Millions of users reported total loss of access or elevated error rates across web interfaces and coding tools.
- Claude (Anthropic): Multiple core models and API endpoints experienced partial to complete outages during the morning hours.
- Gemini (Google): Users reported connection timeouts and failed response generations.
- Microsoft Copilot: Enterprise and individual subscriptions were impacted alongside connected cloud capabilities.
- xAI Grok (Elon Musk): Compute disruptions led to service downtime for xAI users across connected applications.
Having one platform experience temporary downtime is a known challenge in cloud computing. However, having all of these industry leaders suffer simultaneous degradation raised immediate questions about backend dependencies and hardware infrastructure stability.
Detailed Timeline: How the Outage Unfolded and Resolved
Tracking the event from its initial detection to full recovery reveals how quickly incident response teams worked to mitigate the impact across various systems. Here is how the recovery timeline developed throughout the day.
11:00 AM ET — Initial Outage Detection
User error reports began escalating rapidly across user feedback forums and telemetry monitors. DownDetector recorded thousands of simultaneous complaints for OpenAI, Anthropic, Google, Microsoft, and xAI platforms.
Users trying to log into web applications encountered server timeouts, while developers utilizing system APIs noticed high rates of HTTP error codes and failed requests.
This Tweet is currently unavailable. It might be loading or has been removed.
OpenAI updated its status monitor to confirm that the company was actively experiencing issues, highlighting elevated errors across both ChatGPT and Codex services.
At the same time, Anthropic's official status monitor reported widespread disruptions impacting several specific model deployments, including Mythos/Fable 5.1, Mythos/Fable 5, Opus 5, Opus 4.8, and Opus 4.6.
12:24 PM ET — Mitigation and First Signs of Recovery
By early afternoon, user error reports on DownDetector started to decline significantly. Social media users began noting that access to key services was slowly returning to normal in various regions.
OpenAI updated its OpenAI status page to show degraded performance while confirming active interventions. "We have applied the mitigation and are monitoring the recovery," the company stated in an official system notice.
Anthropic also announced major progress on its official status page. The company noted that the issue affecting Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5 was fully resolved, confirming that the impact had ended as of 9:16 AM PT / 16:16 UTC.
1:37 PM ET — Full Restoration Confirmed
By mid-afternoon, all major AI service providers appeared to be back to normal operations, with system traffic stabilizing worldwide.
In a statement provided to Mashable, an Anthropic spokesperson formally confirmed that service was restored across all affected products and developer integrations.
"Claude is fully back up after an infrastructure issue caused a partial outage across Claude.ai, Claude Code, Claude Cowork, and the Claude API earlier today," the Anthropic spokesperson said. "Service was restored at 16:16 UTC."
Thursday Evening Update — SpaceX Statement on Grok and Compute Centers
Later on Thursday evening, SpaceX issued a statement on X (formerly Twitter) regarding the downtime experienced by xAI's Grok chatbot. The company formally attributed the disruption to a localized infrastructure failure.
SpaceX explained that the Grok downtime was caused by "an outage at our Memphis compute center this morning" and offered an official apology to impacted compute partners who relied on the facility's operations.
In a follow-up statement, Elon Musk addressed the situation directly, noting that the company is "taking corrective action to ensure this does not happen again."
This Tweet is currently unavailable. It might be loading or has been removed.
What Caused the Widespread AI Outage?
Understanding what triggers a widespread disruption requires looking closely at how modern machine learning systems are built, hosted, and scaled.
Unlike traditional web applications that consume minimal processing power per query, large language models rely on vast clusters of specialized graphics processing units (GPUs) housed inside high-density compute centers. These high-performance computing centers require complex power networks, intensive cooling systems, and specialized high-speed communication fabrics.
Common Causes of Major AI Outages
When multiple platforms experience downtime at once, engineers typically investigate several core operational factors:
- Data Center Infrastructure Failures: Power grid anomalies, cooling system failures, or hardware breakdowns at major compute centers—such as the Memphis compute facility mentioned by SpaceX—can instantly take entire server clusters offline.
- Shared Cloud Service Dependencies: Many top tech companies rely on the same underlying cloud infrastructure providers, content delivery networks (CDNs), or domain name system (DNS) routing hubs. An issue at a foundational internet layer can ripple across multiple companies simultaneously.
- Network and Traffic Congestion: High-bandwidth data exchange between distributed compute clusters and application backends can create severe network bottlenecks if a single routing path fails.
- System Mitigations and Software Updates: Pushing infrastructure updates, security patches, or load-balancing changes can occasionally trigger cascading unexpected errors across connected API tools like Claude API or OpenAI Codex.
In this specific event, statements from service operators pointed toward distinct compute center infrastructure issues and localized networking problems that were mitigated relatively quickly by engineering teams.
Impact on Developers, Businesses, and Everyday Users
The impact of a **major AI outage** extends far beyond simple consumer inconvenience. Today, thousands of companies rely on integrated artificial intelligence workflows to power customer service bots, automate code writing, process complex datasets, and manage internal operational tasks.
Developer and Enterprise Disruptions
When platforms like Claude API, Claude Code, or OpenAI Codex experience degraded performance, development pipelines around the world hit an immediate bottleneck. Developers who depend on automated code generation or intelligent assistant tools are forced to revert to manual processes.
For enterprises utilizing specialized models like Claude Opus 5 or Claude Mythos 5.1 for business intelligence, unexpected downtime can pause live automated operations, delay software releases, and lower team productivity.
Consumer and Creative Workflows
For individual users, student researchers, and digital creators using web interfaces like ChatGPT or Gemini, service interruptions bring creative tasks to a sudden halt. As artificial intelligence becomes deeply integrated into everyday browser routines, unexpected downtime highlights our growing reliance on external cloud-hosted intelligence.
How to Check ChatGPT Downtime and Claude Status During an Outage
When you encounter connection errors or unresponsiveness while using your favorite tools, you can quickly verify whether the issue is local or part of a broader system failure.
Here are the best ways to monitor platform performance during suspected service failures:
- Check Official Status Pages: Most major providers host real-time incident dashboards. Bookmark the OpenAI status page and the official Claude status page to review active operational alerts and ongoing response updates.
- Consult Third-Party Outage Aggregators: Platforms like DownDetector gather crowd-sourced user reports and automated diagnostic checks to show real-time error spikes across various digital services.
- Review Official Social Media Channels: Engineering teams and company leaders often post rapid updates on platforms like X when major data center or network issues occur.
- Test API Endpoints Independently: If you are a developer, checking raw API responses can help determine whether an issue is isolated to a web front-end interface or affects backend processing engines.
Industry Background and Future Considerations
As the artificial intelligence industry matures, cloud reliability and compute architecture resilience have become top priorities for technology providers. Expanding infrastructure footprint while maintaining 99.9% uptime requires continuous investment in data centers, power management, and redundant network capacity.
At the same time, major technology companies face ongoing legal, operational, and regulatory developments. (Disclosure: Ziff Davis, Mashable's parent company, in April 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
Events like this multi-platform outage serve as a reminder to organizations that building resilient systems requires robust contingency plans. Having backup workflows, multi-provider API fallbacks, and offline redundancy strategies ensures that business operations can continue smoothly even when cloud infrastructure experiences unexpected hiccups.
Final Thoughts on the Recent AI Service Disruptions
While brief isolated glitches happen, experiencing a **major AI outage** across ChatGPT, Claude, Gemini, Copilot, and Grok highlights just how interconnected modern cloud infrastructure truly is. Thanks to swift incident responses and applied mitigations, service across all affected providers was fully restored in short order.
As provider teams implement corrective actions at facility levels like the Memphis compute center, users can expect continued focus on building stronger, more resilient artificial intelligence ecosystems that keep pace with surging global demand.
from Mashable
-via DynaSage
