AI blackout: ChatGPT, Claude and Gemini go dark together — what’s down and what still works?

Users of major artificial intelligence platforms faced widespread disruption on Thursday as OpenAI’s ChatGPT, Anthropic’s Claude and Google’s Gemini reportedly experienced service issues, triggering frustration across social media and disrupting everyday work for developers, students, researchers and office teams.

ChatGPT’s web and mobile apps, its API and Codex are broadly impaired after OpenAI confirmed a major outage. Claude’s chat, API, Claude Code and Claude Cowork are failing across multiple models (Opus 5, Fable 5.1, Mythos 5.1), though Anthropic says a fix is in progress. Gemini users in several countries are seeing intermittent “something went wrong” errors and failed responses, but Google has not confirmed a full outage and some API health checks still show it as operational. Tools that depend on these providers — AI IDEs, no‑code automations, enterprise bots — are also disrupted, while other AI services not tied to these stacks (some local models, smaller providers, and certain Copilot setups) may still be working.

Anthropic confirmed that Claude was experiencing elevated errors across several models and said it had identified the underlying issue and was working on a fix. The incident affected both the Claude consumer service and API, with reports indicating disruption to Claude AI, Claude Code and Claude Cowork.

Users report simultaneous failures

The overlap in reported outages quickly became a talking point on X, formerly Twitter, where users shared screenshots, error messages and jokes about suddenly losing access to their preferred AI tools. Downdetector also showed a sharp rise in user-submitted incident reports, signalling that the problem was affecting more than isolated accounts.

Users expressed a mix of panic and humour as their primary daily workflows came to an abrupt halt.

Claude outage affects users

Claude users reported issues with chat access, coding workflows and API-based tools. Reports on Thursday indicated that the largest concentration of complaints involved Claude Code and Claude Chat, while Anthropic’s status information identified elevated errors across multiple models.

The incident highlights how heavily professionals now rely on generative AI platforms for programming, writing, research, customer support and daily productivity tasks. When several major providers encounter trouble at the same time, users have fewer practical fallback options.

Is ChatGPT down?

OpenAI has previously acknowledged incidents involving elevated latency and partial ChatGPT outages, including cases in which users could not start or continue work in ChatGPT Work. However, simultaneous user reports alone do not establish that one provider’s outage caused another’s.

There is currently no verified evidence that traffic shifting from Claude to ChatGPT directly triggered a ChatGPT outage. High traffic, infrastructure failures, software changes, third-party service dependencies and regional network problems can each cause service disruptions independently.

Google Gemini issues

Google Gemini was also reported by some users as experiencing problems during the broader wave of AI-service complaints. With all three platforms reportedly affected, users searching for alternatives found that switching assistants did not always restore access immediately.

For real-time verification, users should check the official status dashboards for OpenAI, Anthropic and Google rather than relying only on social-media posts or outage-tracker submissions.

The Growing Pain of AI Outages and Dependency

Thursday’s cluster of outages is not an isolated glitch; it is a symptom of how quickly AI has moved from experimental tool to essential infrastructure—and how slowly our systems have adapted to that reality.

From novelty to utility

In just a few years, generative AI has shifted from a curiosity to a core part of daily workflows. Developers use it to write and debug code; students and researchers to summarize papers and brainstorm ideas; office teams to draft emails, reports and presentations; customer-support groups to triage tickets and generate responses. When these services falter, the impact is immediate and visible.

The concentration of usage on a handful of platforms—ChatGPT, Claude, Gemini, Copilot, and a few others—means that when one stumbles, millions feel it. When several stumble together, the sense of disruption multiplies.

Why outages cluster

Several factors explain why AI outages increasingly appear in clusters:

• Shared infrastructure: Major AI providers run on the same hyperscale clouds and global networking layers. A problem in a cloud region, CDN, or edge network can ripple across multiple AI services at once.
• Bursty, heavy workloads: AI requests are computationally expensive and can spike suddenly with new features, model releases, or viral adoption. Systems that scale well under normal load can buckle under sudden surges.
• Tightly coupled stacks: Authentication, billing, logging, model serving, databases, and caching are all interdependent. A failure in one layer—such as a database connection pool exhausting during scaling—can cascade into a full API outage.
• Human error under pressure: Many large incidents begin with a configuration change, deployment, or scaling operation that behaves unexpectedly under real-world load.

These dynamics do not excuse outages, but they help explain why “one after another” has become a familiar pattern.

How dependent are we?

The dependency is deeper than most users realize:

• Enterprises have embedded AI into internal tools, support bots and analytics pipelines.
• Developers rely on AI coding assistants integrated into IDEs and CI/CD workflows.
• No-code and low-code builders use AI for automations, forms and customer-facing experiences.

Consumers use AI for everyday tasks: drafting messages, planning trips, learning new skills.

When AI goes down, it is not just a chatbot that stops working. Entire workflows stall, deadlines slip, and teams lose their primary way of getting things done. The risk is no longer theoretical; it is operational and financial.

What this means going forward

We are in a transition period where AI is becoming critical infrastructure faster than our resilience practices can keep up. That gap is what Thursday’s outages exposed.

For individuals and organizations that depend on AI, the lesson is clear: treat AI providers like any other essential service. Map your dependencies, monitor not just uptime but latency and error rates, design graceful degradation and fallbacks, and practice what happens when your main AI provider is unavailable for hours.

AI is not going away. But as it becomes more central to how we work and live, the cost of taking its reliability for granted will only grow.