AnthropicのFrontier Red TeamとMozillaがAIを用いた脆弱(ぜいじゃく)性検出に関する連携を行い、Claude Opus 4.6がわずか2週間の調査でFirefoxについて計112件の報告を提出し、その中から22件の脆弱性が確認されたことを報告しました。この成果は、AIが大規模なコードベースの安全性を極めて高い速度で検証し、強化できる可能性を実証するものです。

In a recent security partnership with Mozilla, Anthropic found 22 separate vulnerabilities in Firefox — fourteen of them ...

A high-severity Chrome vulnerability has allowed malicious extensions to exploit the Gemini panel and gain elevated access to ...

The proposed WebAssembly Component Model would improve web integration and elevate Wasm from “second-class language” status, says Mozilla developer.

Builderius page builder announced an experimental AI integration that can read and apply changes directly inside the builder.