{
  "id": "anthropic-sitemap:research:prompt-injection-defenses",
  "type": "article",
  "title": "Mitigating the risk of prompt injections in browser use",
  "abstract": "Claude Opus 4.5 sets a new standard in robustness to _prompt injections_—adversarial instructions hidden within the content that AI models process. Our new model is a major improvement over previous ones in both its core performance and in the safeguards surrounding its use. But prompt injection is far from a solved problem, particularly as models take more real-world actions. We expect to continue our progress—aiming for a future where AI models (or \"agents\") can handle high-value tasks without significant prompt injection risk.",
  "issued": {
    "date-parts": [
      [
        2025,
        11,
        24
      ]
    ]
  },
  "URL": "https://www.anthropic.com/research/prompt-injection-defenses",
  "publisher": "Anthropic",
  "source": "vendor/anthropic-sitemap/research/prompt-injection-defenses.md"
}