- Claude for Teachers offers free premium access to US K-12 educators with standards-aligned lesson planning and differentiation tools.
- Claude Science provides scientists with a specialized toolkit for research workflows, though early users note accuracy and biosecurity guardrail issues.
- Claude Code receives a 50% usage limit promotion through July 19, 2026, while comparisons highlight its strength in code understanding versus Codex's speed.
- A security vulnerability (ClaudeBleed) in the Claude for Chrome extension allows malicious extensions to hijack commands, prompting user caution.
Anthropic has been busy rolling out new specialized versions of its Claude AI assistant, targeting educators, scientists, and developers. The company also faced a security scare with its Chrome extension, while simultaneously offering a time-limited boost for its coding tool. Here’s a roundup of the latest developments.
The past few weeks have seen Claude expand into classrooms and labs, with tools designed to save time and improve outcomes. However, not everything has been smooth sailing—a vulnerability in the browser extension raised eyebrows, and early feedback on the science toolkit highlights both promise and pitfalls.
Claude for Teachers: Free Access for US Educators
Anthropic launched Claude for Teachers, a free tool for verified K-12 educators in the US. It provides premium Claude capabilities, a library of teaching skills, and a direct connection to evidence-based curricula mapped to academic standards in all 50 states. The tool is designed to close the gap between educational best practices and what a teacher’s week allows, drawing on decades of research showing that differentiation, mastery-based learning, and small group instruction improve student achievement. Early evidence suggests AI tools for teachers can strengthen instructional practice and improve student outcomes, unlike mixed results for student-facing AI. Claude for Teachers connects to Learning Commons, giving access to academic standards and trusted resources like OpenSciEd and Illustrative Mathematics. Educators can also integrate with tools like ASSISTments, Brisk Teaching, Canva Education, and others. The offering includes Claude Code and Cowork for tasks like analyzing class data and scheduling repeated tasks. Data is not used for model training, and student information is protected under a K-12 Data Processing Addendum compliant with FERPA. Anthropic is working with the American Federation of Teachers to align with gold standards for safety and privacy. Additionally, a free AI Fluency course co-created with Teach For America and the AFT is available. Educators can sign up by June 30, 2027, for a full year of access.
Claude Science: A New Research Assistant
Anthropic also introduced Claude Science, a toolkit integrating tools and packages scientists use into a single agentic offering. It can analyze and produce code, generate images, and suggest workflows for wet lab work, with over 60 pre-configured connectors for genomics, proteomics, and structural biology. Early users like neuroscientist Stephen Francis and Allen Institute’s Jerome Lecoq reported faster literature reviews and coding support, though they noted occasional mistakes and inability to access paywalled papers. The tool is more transparent than regular Claude, showing step-by-step reasoning and running more analysis locally for privacy. However, some researchers found biosecurity guardrails too stifling, and the model still struggles with critiquing data quality like a human scientist. All three scientists said they would continue using it, but raised concerns about the impact on hiring and the scientific model.
Claude Code: Promotion and Performance
Anthropic extended a limited-time promotion increasing Claude Code weekly usage limits by 50% through July 19, 2026, for Pro, Max, and Team plans. The boost applies only to Claude Code across CLI, IDE extensions, desktop app, and web. Meanwhile, comparisons between Claude Code, OpenAI Codex, and OpenCode show that Claude Code excels at deep codebase understanding and careful multi-file refactors, while Codex is faster for well-specified tasks and OpenCode offers model freedom and LSP integration. A separate opinion piece noted that Codex is strong for backend work but produces bland frontend designs, whereas Claude Code delivers more aesthetically appealing UIs. The verdict: Claude Code is best for unfamiliar codebases, Codex for throughput, and OpenCode for control and privacy. Users are advised to keep their hand on the wheel as all tools can produce plausible but wrong code.
Security Alert: ClaudeBleed Vulnerability
A security issue dubbed ClaudeBleed was reported in the Claude for Chrome extension. It allows a malicious browser extension to pretend to be Claude’s website and secretly drive the extension to read data and take actions in user accounts, such as reading Gmail or sending emails. Anthropic acknowledged the reports and closed them, but researchers claim the fundamental privilege handoff remains brittle—a bypass is still possible with six lines of JavaScript. Users are advised to turn off “Act without asking,” review extensions, limit sensitive account access, and consider disabling the extension on systems handling sensitive data until a more comprehensive fix is shipped.
Anthropic’s Claude continues to evolve across multiple domains, offering powerful tools for teachers, scientists, and developers. While the security issue serves as a reminder to stay cautious, the company’s rapid iteration and partnerships suggest a strong commitment to making AI useful in real-world settings. Users should weigh the benefits against the risks, especially when granting browser permissions.
