Anthropic pushes enterprise
- Anthropic released Claude Design in research preview for Pro, Max, Team and Enterprise plans while noting admin gaps for audits and usage tracking. - Its Mythos model reportedly found 271 Firefox vulnerabilities, showing use in security workflows but also raising governance questions. - The rollout shows enterprise demand colliding with missing controls and pricing tests as Anthropic scales commercially ( ).
Anthropic has started pushing Claude deeper into workplaces with a new design tool, even as its own help pages say key admin controls are not there yet. (support.claude.com) Claude Design entered research preview on April 17 alongside Opus 4.7 and is available on Pro, Max, Team, and Enterprise plans. Anthropic says the product lets users generate designs, interactive prototypes, presentations, and one-pagers through a chat window linked to a visual canvas. (support.claude.com, support.claude.com) For Team and Enterprise accounts, Claude Design can ingest codebases, slide decks, logos, and brand guidelines, then turn them into a shared design system that all projects inherit. Enterprise access is off by default, according to Anthropic’s setup guides. (support.claude.com, support.claude.com) Anthropic’s pricing pages show why the launch looks like a commercial test as much as a product release. Claude Design has its own weekly allowances, separate metering from chat and Claude Code, and paid overages; usage-based Enterprise customers get a one-time credit that Anthropic says covers about 20 typical prompts and expires July 17. (support.claude.com) The same pricing page says Claude Design “doesn’t support audit logs or usage tracking yet.” That caveat lands awkwardly beside Anthropic’s August 2025 push for business customers, when it promoted Compliance API access, usage analytics, spend caps, and other admin controls for Claude Code and enterprise plans. (support.claude.com, anthropic.com) At the same time, Anthropic is pitching Claude for security work that carries higher stakes than slide decks. SecurityWeek reported on April 22 that Mozilla said an early Claude Mythos Preview found 271 vulnerabilities in Firefox, with fixes shipping this week in Firefox 150. (securityweek.com) Mozilla’s public advisory credited Claude with three Common Vulnerabilities and Exposures, or CVEs, while SecurityWeek said the larger 271 total likely included lower-severity bugs and hardening issues that do not get public CVE numbers. Firefox Chief Technology Officer Bobby Holley told the outlet the bugs were still within the range of what “an elite human researcher” could find. (securityweek.com) SecurityWeek also said Anthropic is limiting Mythos distribution through Project Glasswing rather than releasing it broadly, with participants including Amazon Web Services, Apple, Cisco, Google, JPMorgan Chase, Microsoft, Nvidia, and Palo Alto Networks. Anthropic’s own sales language for business plans has emphasized centralized billing, role-based access, and predictable spend as it tries to win larger deployments. (securityweek.com, anthropic.com) The result is a familiar enterprise pattern: Anthropic is expanding Claude from chat into design and security workflows, while some of the controls companies usually want are still arriving product by product. For now, the company is asking customers to adopt new capabilities first and wait for parts of the governance stack to catch up. (support.claude.com, anthropic.com)
