Anthropic Just Pulled Two AI Models Worldwide — What It Means for UAE Users

UAE developers woke up to find their latest AI tools suddenly unavailable. Anthropic has disabled access to its newly released Fable 5 and Mythos 5 models worldwide following a US government directive, as reported by PCMag. While the order specifically targeted foreign nationals over national security concerns, Anthropic extended the ban globally, affecting developers from Dubai to Abu Dhabi who depend on these models for coding and cybersecurity work.

What happened and why

The shutdown stems from what the US government considers a security vulnerability. Officials discovered a 'narrow' jailbreak method for Fable 5 that reportedly involves feeding code into the model for review — a common task for developers — and then exploiting it. Mythos 5, known internally as Project Glasswing, was already limited to select US partners due to its ability to identify software vulnerabilities in major platforms like Mozilla Firefox and potentially Apple's macOS.

Anthropic pushed back against the decision, stating it 'disagree[s] that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.' The company has faced regulatory pressure before — the Pentagon labelled it a 'supply-chain risk' in May, and President Trump called the company 'left-wing nutjobs.'

Which Claude models still work in UAE

The good news is that older Anthropic models remain unaffected. Claude Opus 4.8, the previous flagship, continues to operate normally and handles most enterprise tasks for which Fable 5 was designed. If your workflow was built around Fable 5's enhanced coding capabilities, you'll need to roll back API calls to the Opus 4.8 tier.

Claude Pro subscriptions remain available at approximately AED 75 per month, though enterprise API pricing varies based on token usage. The transition shouldn't disrupt most business operations, though you may notice reduced performance on complex coding tasks that Fable 5 handled better.

What UAE developers should do now

This incident exposes the risks of depending entirely on US-based AI infrastructure. For GCC-based CTOs, it suggests Anthropic may face more regulatory hurdles than competitors like OpenAI or Google.

Immediate actions:

• Switch API endpoints: Update any applications calling Fable 5 to use Claude Opus 4.8 instead
• Review code analysis workflows: The jailbreak concerns specifically relate to how these models handle code review tasks
• Diversify your AI stack: Ensure applications can failover to models like GPT-4o or other alternatives
• Assess regulatory risk: Factor potential US government intervention into your AI vendor selection

The broader GCC impact

Microsoft separately blocked its staff from using Claude Fable 5 this week over data retention concerns, showing how quickly enterprise policies can shift. The UAE's ambitions to become a global AI hub highlight the vulnerability of relying on foreign infrastructure subject to geopolitical pressures.

The timing is particularly notable as AI models increasingly handle coding tasks that were human-only territory just months ago. When a single government order can pull commercial AI tools from around the world overnight, it underscores the need for more diverse supplier relationships.

Current Claude pricing and availability

Claude Opus 4.8 remains fully available through Anthropic's website for UAE users. Claude Pro costs approximately AED 75 per month, while enterprise pricing depends on API token usage. No timeline has been provided for when — or if — Fable 5 and Mythos 5 will return, as the shutdown resulted from a government directive rather than technical issues.

Written by

Share

Copied