Anthropic’s Claude Code changed the AI game last year when it turbocharged our ability to vibe code. No longer did you need a comprehensive understanding of coding languages to build new apps, websites and widgets; you could describe what you want in plain words, and the AI assistant would make it happen. Now, it’s getting into more creative work.

An update to Claude Design, announced on Wednesday, is intended to make it easier for Claude users to integrate design and visual assets into their work. Claude Design launched in beta in April, and this week’s updates make it easier to keep your AI content aligned with your company’s brand guidelines.

There will be a new administrator role; this person can set up your “design system,” which is essentially your brand kit that people can use across projects, not just haphazardly. Claude will automatically check for compliance. You can now import your GitHub repo, design files and raw uploads to your design system.

Claude Code will now be able to pull from your designs and vice versa. You can bring your designs into your coding terminal with the command “/design.” The two capabilities are synced, so you should be able to pull from the most up-to-date elements from each — no back-and-forth needed.

It’s been a busy week for Anthropic, and it’s only Wednesday. The company was forced to pull its newest AI model, Fable 5, after the US government ordered the company to prevent non-US citizens from using it. The export control order was issued over the weekend after it found a “jailbreak” that circumvented the model’s cybersecurity guardrails. The only way to comply with the order, Anthropic said, was to pull the model for everyone, as CNET reported this week.

The debacle is the latest chapter in a growing conflict between the US government and AI developers. Anthropic very publicly refused to let the Department of Defense use Claude in certain cases involving surveillance and fully autonomous weapons. OpenAI, one of Anthropic’s biggest competitors that makes ChatGPT, was quick to fill the void and snatch up a Pentagon contract. But questions about the role of AI in military operations are still largely unanswered.

Anthropic was also sued this week by a group of Claude subscribers alleging that the company lied about usage limits for its paid plans.