{"id":25865,"date":"2025-12-26T11:24:49","date_gmt":"2025-12-26T11:24:49","guid":{"rendered":"https:\/\/itbusinesstoday.com\/?p=25865"},"modified":"2025-12-29T13:19:13","modified_gmt":"2025-12-29T13:19:13","slug":"ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection","status":"publish","type":"post","link":"https:\/\/itbusinesstoday.com\/ja\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/","title":{"rendered":"\u30ea\u30b3\u30fc\u3001\u6709\u5bb3\u306aLLM\u51fa\u529b\u3092\u691c\u77e5\u3059\u308b\u65b0\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u30e2\u30c7\u30eb\u3067AI\u751f\u6210\u5b89\u5168\u6027\u3092\u5f37\u5316"},"content":{"rendered":"

12\u670825\u65e5\u3001\u30ea\u30b3\u30fc\u306f\u3001\u65b0\u958b\u767a\u306e \u201c<\/strong>\u30bb\u30fc\u30d5\u30ac\u30fc\u30c9\u30e2\u30c7\u30eb\u201d<\/strong> \u306f\u3001\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM\uff09\u306b\u3088\u3063\u3066\u751f\u6210\u3055\u308c\u305f\u6709\u5bb3\u306a\u60c5\u5831\u3092\u691c\u51fa\u3059\u308b\u305f\u3081\u306b\u8a2d\u8a08\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u4eca\u56de\u306e\u30a2\u30c3\u30d7\u30c7\u30fc\u30c8\u306f\u3001\u5f93\u6765\u306e\u5165\u529b\u30d5\u30a3\u30eb\u30bf\u30ea\u30f3\u30b0\u306b\u3068\u3069\u307e\u3089\u305a\u3001\u554f\u984c\u306e\u3042\u308bLLM\u306e\u51fa\u529b\u3092\u76e3\u8996\u3057\u3066\u30d6\u30ed\u30c3\u30af\u3059\u308b\u3053\u3068\u3067\u3001\u30a8\u30f3\u30bf\u30fc\u30d7\u30e9\u30a4\u30baAI\u5c0e\u5165\u306e\u305f\u3081\u306e\u591a\u5c64\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u30fb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u69cb\u7bc9\u3057\u3001\u30b8\u30a7\u30cd\u30ec\u30fc\u30c6\u30a3\u30d6AI\u306e\u5b89\u5168\u6027\u3092\u5927\u304d\u304f\u9032\u5316\u3055\u305b\u307e\u3059\u3002.<\/p>

As generative AI grows in use across industries, worries about misinformation, data leaks, discrimination, and inappropriate content have increased. Ricoh’s announcement tackles these risks head-on. There’s a rising demand from companies, especially in Japan, for AI systems. They want systems that are not just powerful but also trustworthy, auditable, and in line with business and regulatory needs.<\/p>

\u30d7\u30ed\u30f3\u30d7\u30c8\u30fb\u30d5\u30a3\u30eb\u30bf\u30ea\u30f3\u30b0\u304b\u3089\u30d5\u30eb\u30b9\u30bf\u30c3\u30afAI\u30bb\u30fc\u30d5\u30c6\u30a3\u3078<\/strong><\/h3>

\u6700\u8fd1\u307e\u3067\u3001\u307b\u3068\u3093\u3069\u306e\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u30fb\u30b7\u30b9\u30c6\u30e0\u306f\u3001\u6b21\u306e\u3088\u3046\u306a\u70b9\u306b\u91cd\u70b9\u3092\u7f6e\u3044\u3066\u3044\u307e\u3057\u305f\u3002 \u6709\u5bb3\u306a\u30d7\u30ed\u30f3\u30d7\u30c8<\/strong> LLM\u306b\u3088\u3063\u3066\u51e6\u7406\u3055\u308c\u308b\u524d\u306b\u3002\u3057\u304b\u3057\u3001\u3053\u306e\u30a2\u30d7\u30ed\u30fc\u30c1\u306b\u306f\u660e\u3089\u304b\u306a\u9650\u754c\u304c\u3042\u308a\u307e\u3059\u3002\u5b89\u5168\u306a\u5165\u529b\u3067\u3042\u3063\u3066\u3082\u3001\u9ad8\u5ea6\u306a\u30e2\u30c7\u30eb\u3067\u306f\u3001\u5e7b\u899a\u3084\u6587\u8108\u306e\u8aa4\u89e3\u3001\u3042\u308b\u3044\u306f\u610f\u56f3\u3057\u306a\u3044\u63a8\u8ad6\u7d4c\u8def\u306b\u3088\u3063\u3066\u3001\u554f\u984c\u306e\u3042\u308b\u51fa\u529b\u3092\u751f\u6210\u3059\u308b\u53ef\u80fd\u6027\u304c\u3042\u308a\u307e\u3059\u3002.<\/p>

Ricoh’s new Safeguard Model tackles this gap by monitoring both the input prompts and the responses generated by the LLM. This dual-layer detection allows organizations to intervene not only before execution, but also after content generation, significantly reducing the risk of harmful information being exposed to users or downstream systems.<\/p>

\u3053\u306e\u958b\u767a\u306f Ricoh’s<\/a> 2024\u5e7410\u6708\u3001LLM\u306e\u793e\u5185\u5b89\u5168\u6d3b\u52d5\u3092\u958b\u59cb\u30022025\u5e748\u6708\u306b\u306f\u6709\u5bb3\u30d7\u30ed\u30f3\u30d7\u30c8\u691c\u51fa\u6a5f\u80fd\u3092\u30ea\u30ea\u30fc\u30b9\u3057\u3001RICOH\u30aa\u30f3\u30d7\u30ec\u30df\u30b9LLM\u30b9\u30bf\u30fc\u30bf\u30fc\u30ad\u30c3\u30c8\u306b\u6a19\u6e96\u642d\u8f09\u3002\u4eca\u56de\u767a\u8868\u3055\u308c\u305f\u51fa\u529b\u691c\u51fa\u6a5f\u80fd\u306b\u3088\u308a\u3001\u3053\u306e\u5b89\u5168\u30eb\u30fc\u30d7\u304c\u5b8c\u6210\u3057\u3001\u30a8\u30f3\u30bf\u30fc\u30d7\u30e9\u30a4\u30ba\u5411\u3051\u751f\u6210AI\u306e\u5805\u7262\u306a\u6df1\u5c64\u9632\u5fa1\u30e2\u30c7\u30eb\u306b\u8fd1\u3065\u304d\u307e\u3059\u3002.<\/p>

\u3053\u3061\u3089\u3082\u304a\u8aad\u307f\u304f\u3060\u3055\u3044\uff1a Digital Securities Completes ¥300 Million Series A Round, Bringing Total Funding to ¥1.5 Billion<\/a><\/strong><\/h4>

\u6280\u8853\u632f\u8208\u8ca1\u56e3\u3068\u65e5\u672c\u8a9e\u6700\u9069\u5316<\/strong><\/h3>

Safeguard Model\u306f\u3001Meta\u793e\u304c\u958b\u767a\u3057\u305fMeta-Llama-3.1-8B\u306e\u6539\u826f\u7248\u3067\u3042\u308bLlama-3.1-Swallow-8B-Instruct-v0.5\u3092\u30d9\u30fc\u30b9\u306b\u3057\u3066\u3044\u307e\u3059\u3002\u30ea\u30b3\u30fc\u306f\u3001\u56fd\u5185\u306e\u4f01\u696d\u74b0\u5883\u3067\u6b63\u78ba\u306a\u30b3\u30f3\u30c6\u30f3\u30c4\u5206\u985e\u3092\u884c\u3046\u305f\u3081\u306b\u91cd\u8981\u306a\u8981\u4ef6\u3067\u3042\u308b\u65e5\u672c\u8a9e\u306e\u7406\u89e3\u5ea6\u3092\u5411\u4e0a\u3055\u305b\u308b\u305f\u3081\u306b\u3001\u3053\u306e\u57fa\u76e4\u3092\u9078\u629e\u3057\u307e\u3057\u305f\u3002.<\/p>

\u65e5\u672c\u8a9e\u306e\u30cb\u30e5\u30a2\u30f3\u30b9\u306b\u5408\u308f\u305b\u3066\u30e2\u30c7\u30eb\u3092\u5fae\u8abf\u6574\u3059\u308b\u3053\u3068\u3067\u3001\u30bb\u30fc\u30d5\u30ac\u30fc\u30c9\u30b7\u30b9\u30c6\u30e0\u306f\u6587\u8108\u3001\u30c8\u30fc\u30f3\u3001\u610f\u56f3\u3092\u7406\u89e3\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002\u3053\u308c\u3089\u306e\u8981\u7d20\u306f\u3001\u30eb\u30fc\u30eb\u30d9\u30fc\u30b9\u306e\u30b7\u30b9\u30c6\u30e0\u304c\u898b\u843d\u3068\u3057\u304c\u3061\u306a\u5fae\u5999\u306a\u6709\u5bb3\u30b3\u30f3\u30c6\u30f3\u30c4\u3092\u767a\u898b\u3059\u308b\u305f\u3081\u306e\u9375\u3068\u306a\u308a\u307e\u3059\u3002.<\/p>

\u30e2\u30c7\u30eb\u306f\u4f55\u5343\u3082\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3067\u5b66\u7fd2\u3055\u308c\u307e\u3057\u305f\u3002\u3053\u308c\u3089\u306f14\u306e\u30ea\u30b9\u30af\u30e9\u30d9\u30eb\u306b\u5206\u985e\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u305d\u306e\u4e2d\u306b\u306f\u66b4\u529b\u3001\u72af\u7f6a\u3001\u5dee\u5225\u3001\u30d7\u30e9\u30a4\u30d0\u30b7\u30fc\u4fb5\u5bb3\u3001\u305d\u306e\u4ed6\u306e\u30c7\u30ea\u30b1\u30fc\u30c8\u306a\u30c8\u30d4\u30c3\u30af\u304c\u542b\u307e\u308c\u3066\u3044\u307e\u3059\u3002\u3053\u306e\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306f\u3001\u30b7\u30b9\u30c6\u30e0\u304c\u660e\u78ba\u306a\u30ea\u30b9\u30af\u3068\u96a0\u308c\u305f\u30ea\u30b9\u30af\u306e\u4e21\u65b9\u3092\u767a\u898b\u3059\u308b\u306e\u306b\u5f79\u7acb\u3061\u307e\u3059\u3002\u6709\u5bb3\u306a\u30d7\u30ed\u30f3\u30d7\u30c8\u3084\u5fdc\u7b54\u304c\u4f7f\u7528\u3055\u308c\u305f\u308a\u5171\u6709\u3055\u308c\u305f\u308a\u3059\u308b\u524d\u306b\u30d6\u30ed\u30c3\u30af\u3057\u307e\u3059\u3002.<\/p>

\u30ea\u30b3\u30fc\u306b\u3088\u308b\u3068\u3001\u3053\u306e\u30e2\u30c7\u30eb\u306e\u6027\u80fd\u8a55\u4fa1\u3067\u306f\u3001\u4ed6\u306e\u30d9\u30f3\u30c0\u30fc\u306e\u540c\u7b49\u306e\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u30bd\u30ea\u30e5\u30fc\u30b7\u30e7\u30f3\u3088\u308a\u3082\u9ad8\u3044\u691c\u51fa\u7cbe\u5ea6\u3092\u9054\u6210\u3057\u3066\u304a\u308a\u3001\u3053\u306e\u30a2\u30d7\u30ed\u30fc\u30c1\u304c\u6210\u719f\u3057\u3066\u3044\u308b\u3053\u3068\u3092\u88cf\u4ed8\u3051\u3066\u3044\u307e\u3059\u3002.<\/p>

\u65e5\u672c\u306e\u30c6\u30af\u30ce\u30ed\u30b8\u30fc\u7523\u696d\u3078\u306e\u5f71\u97ff<\/strong><\/h3>

\u8cac\u4efb\u3042\u308bAI\u306e\u57fa\u6e96\u3092\u9ad8\u3081\u308b<\/strong><\/h4>

Ricoh’s announcement shows a bigger change in Japan’s tech industry. It’s moving toward responsible, enterprise-level AI use. As companies shift generative AI from testing to main operations, safety measures like output guardrails will be crucial. They will no longer be optional but rather essential infrastructure.<\/p>

\u3053\u306e\u5909\u5316\u306f\u3001\u4ed6\u306e\u65e5\u672c\u306e\u30cf\u30a4\u30c6\u30af\u4f01\u696d\u306b\u3082\u5f71\u97ff\u3092\u4e0e\u3048\u308b\u53ef\u80fd\u6027\u304c\u3042\u308a\u307e\u3059\u3002\u7279\u306b\u3001LLM\u30d7\u30e9\u30c3\u30c8\u30d5\u30a9\u30fc\u30e0\u3001AI\u30b5\u30fc\u30d3\u30b9\u3001\u307e\u305f\u306f\u696d\u754c\u306b\u7279\u5316\u3057\u305fAI\u30c4\u30fc\u30eb\u3092\u6301\u3064\u4f01\u696d\u306b\u5f53\u3066\u306f\u307e\u308a\u307e\u3059\u3002\u540c\u69d8\u306e\u591a\u5c64\u5b89\u5168\u8a2d\u8a08\u3092\u63a1\u7528\u3057\u59cb\u3081\u308b\u304b\u3082\u3057\u308c\u307e\u305b\u3093\u3002\u6642\u9593\u306e\u7d4c\u904e\u3068\u3068\u3082\u306b\u3001\u51fa\u529b\u3092\u610f\u8b58\u3057\u305f\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u306f\u3001\u898f\u5236\u3055\u308c\u305f\u5206\u91ce\u3084\u30ea\u30b9\u30af\u306b\u654f\u611f\u306a\u5206\u91ce\u3067\u8ca9\u58f2\u3055\u308c\u308bAI\u88fd\u54c1\u306e\u57fa\u672c\u7684\u306a\u671f\u5f85\u306b\u306a\u308b\u304b\u3082\u3057\u308c\u307e\u305b\u3093\u3002.<\/p>

Alignment with Japan’s Governance-First AI Culture<\/strong><\/h4>

Japan has consistently emphasized governance, trust, and risk management in its approach to emerging technologies. Ricoh’s on-premise, customizable safeguard model aligns well with this philosophy, offering organizations control over how AI behaves within their specific business and compliance contexts.<\/p>

This approach contrasts with one-size-fits-all cloud moderation systems and may strengthen Japan’s position as a leader in enterprise-safe generative AI architectures.<\/p>

\u65e5\u672c\u306b\u304a\u3051\u308b\u4e8b\u696d\u3078\u306e\u5f71\u97ff<\/strong><\/h3>

\u4f01\u696d\u306b\u3068\u3063\u3066\u3001\u6709\u5bb3\u306aAI\u306e\u51fa\u529b\u3092\u691c\u51fa\u3057\u3001\u30d6\u30ed\u30c3\u30af\u3059\u308b\u80fd\u529b\u306f\u3001\u76f4\u63a5\u7684\u306a\u696d\u52d9\u4e0a\u306e\u4fa1\u5024\u3068\u98a8\u8a55\u4e0a\u306e\u4fa1\u5024\u304c\u3042\u308a\u307e\u3059\u3002.<\/p>

\u6cd5\u7684\u30ea\u30b9\u30af\u3068\u98a8\u8a55\u30ea\u30b9\u30af\u306e\u4f4e\u6e1b<\/strong><\/h4>

\u91d1\u878d\u3001\u533b\u7642\u3001\u88fd\u9020\u3001\u516c\u5171\u30b5\u30fc\u30d3\u30b9\u306a\u3069\u306e\u5206\u91ce\u3067\u306f\u3001\u4e0d\u9069\u5207\u306aAI\u51fa\u529b\u304c1\u4ef6\u3067\u3082\u767a\u751f\u3059\u308b\u3068\u3001\u30b3\u30f3\u30d7\u30e9\u30a4\u30a2\u30f3\u30b9\u9055\u53cd\u3084\u4fe1\u983c\u306e\u5931\u589c\u306b\u3064\u306a\u304c\u308b\u53ef\u80fd\u6027\u304c\u3042\u308a\u307e\u3059\u3002\u51fa\u529b\u30ec\u30d9\u30eb\u306e\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u306f\u3001AI\u304c\u751f\u6210\u3059\u308b\u30b3\u30f3\u30c6\u30f3\u30c4\u304c\u4f01\u696d\u30dd\u30ea\u30b7\u30fc\u3084\u793e\u4f1a\u898f\u7bc4\u306b\u6cbf\u3063\u305f\u3082\u306e\u3067\u3042\u308b\u3053\u3068\u3092\u4fdd\u8a3c\u3059\u308b\u3053\u3068\u3067\u3001\u3053\u306e\u3088\u3046\u306a\u30ea\u30b9\u30af\u3092\u4f4e\u6e1b\u3057\u307e\u3059\u3002.<\/p>

\u4fe1\u983c\u6027\u306e\u9ad8\u3044AI\u306e\u8fc5\u901f\u306a\u5c0e\u5165<\/strong><\/h4>

One of the biggest barriers to generative AI adoption has been fear of unintended consequences. By embedding safety at both the input and output layers, Ricoh’s solution lowers the risk threshold, enabling organizations to deploy AI more broadly across internal workflows, customer support, and knowledge management.<\/p>

\u30d3\u30b8\u30cd\u30b9\u30fb\u30b3\u30f3\u30c6\u30af\u30b9\u30c8\u306e\u30ab\u30b9\u30bf\u30de\u30a4\u30ba<\/strong><\/h4>

\u30ea\u30b3\u30fc\u306f\u3001\u696d\u52d9\u306b\u95a2\u4fc2\u306e\u306a\u3044\u30b3\u30f3\u30c6\u30f3\u30c4\u3092\u30d6\u30ed\u30c3\u30af\u3059\u308b\u306a\u3069\u3001\u4e00\u822c\u7684\u306a\u88ab\u5bb3\u9632\u6b62\u306b\u3068\u3069\u307e\u3089\u306a\u3044\u30ab\u30b9\u30bf\u30de\u30a4\u30ba\u3092\u53ef\u80fd\u306b\u3059\u308b\u8a08\u753b\u3092\u793a\u3057\u3066\u3044\u307e\u3059\u3002\u6587\u8108\u3092\u8003\u616e\u3057\u305fAI\u3078\u306e\u6249\u3092\u958b\u304d\u307e\u3059 \u30ac\u30d0\u30ca\u30f3\u30b9<\/strong>, \u30e2\u30c7\u30eb\u3067\u306f\u3001\u666e\u904d\u7684\u306a\u5b89\u5168\u30eb\u30fc\u30eb\u3060\u3051\u3067\u306a\u304f\u3001\u7d44\u7e54\u306e\u76ee\u7684\u3084\u751f\u7523\u6027\u306e\u76ee\u6a19\u306b\u3088\u3063\u3066\u3082\u5236\u7d04\u3092\u53d7\u3051\u307e\u3059\u3002.<\/p>

\u4eca\u5f8c\u306e\u5c55\u671b<\/strong><\/h3>

\u30ea\u30b3\u30fc\u306f\u3001\u65b0\u3057\u3044\u30bb\u30fc\u30d5\u30ac\u30fc\u30c9\u30e2\u30c7\u30eb\u3092RICOH\u30aa\u30f3\u30d7\u30ec\u30df\u30b9LLM\u30b9\u30bf\u30fc\u30bf\u30fc\u30ad\u30c3\u30c8\u306b\u6a19\u6e96\u642d\u8f09\u3057\u307e\u3059\u3002\u3053\u308c\u306f\u3001\u30bb\u30ad\u30e5\u30a2\u3067\u30a8\u30f3\u30bf\u30fc\u30d7\u30e9\u30a4\u30ba\u5bfe\u5fdc\u306e\u30b8\u30a7\u30cd\u30ec\u30fc\u30c6\u30a3\u30d6AI\u306b\u5bfe\u3059\u308b\u30ea\u30b3\u30fc\u306e\u30b3\u30df\u30c3\u30c8\u30e1\u30f3\u30c8\u3092\u793a\u3059\u3082\u306e\u3067\u3059\u3002\u8aac\u660e\u53ef\u80fd\u3067\u3001\u5236\u5fa1\u53ef\u80fd\u3067\u3001\u30b3\u30f3\u30d7\u30e9\u30a4\u30a2\u30f3\u30b9\u306b\u6e96\u62e0\u3057\u305fAI\u3078\u306e\u9700\u8981\u304c\u9ad8\u307e\u308b\u4e2d\u3001\u30a2\u30a6\u30c8\u30d7\u30c3\u30c8\u3092\u610f\u8b58\u3057\u305f\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u306f\u3001AI\u5c0e\u5165\u306e\u6b21\u306e\u6bb5\u968e\u3092\u5f62\u4f5c\u308b\u3053\u3068\u306b\u306a\u308b\u3067\u3057\u3087\u3046\u3002.<\/p>

In a market where trust matters as much as performance, Ricoh’s move shows an important truth. The future of generative AI will depend not just on what models create, but also on how safely and responsibly they do it.<\/p>","protected":false},"excerpt":{"rendered":"

On December 25, Ricoh announced the development of a new “Safeguard Model” designed to detect harmful information generated by large language models (LLMs). The update marks a significant evolution in generative AI safety, extending beyond traditional input filtering to also monitor and block problematic LLM outputs, creating a multi-layered guardrail architecture for enterprise AI deployments.As generative AI grows in use across industries, worries about misinformation, data leaks, discrimination, and inappropriate content have increased. Ricoh’s announcement tackles these risks head-on. There’s a rising demand from companies, especially in Japan, for AI systems. They want systems that are not just powerful but also trustworthy, auditable, and in line with business and regulatory […]<\/p>\n","protected":false},"author":7,"featured_media":25920,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_bbp_topic_count":0,"_bbp_reply_count":0,"_bbp_total_topic_count":0,"_bbp_total_reply_count":0,"_bbp_voice_count":0,"_bbp_anonymous_reply_count":0,"_bbp_topic_count_hidden":0,"_bbp_reply_count_hidden":0,"_bbp_forum_subforum_count":0,"wprm-recipe-roundup-name":"","wprm-recipe-roundup-description":"","postBodyCss":"","postBodyMargin":[],"postBodyPadding":[],"postBodyBackground":{"backgroundType":"classic","gradient":""},"footnotes":""},"categories":[184,8098,85],"tags":[10284,10283,105,10282,106,10281],"ppma_author":[572],"class_list":["post-25865","post","type-post","status-publish","format-standard","has-post-thumbnail","category-cybersecurity","category-news-articles","category-tech","tag-guardrail-architecture","tag-guardrail-model","tag-itbusinesstoday","tag-llm-output-detection","tag-news","tag-safeguard-model"],"yoast_head":"\nRicoh Strengthens AI Safety with Guardrail for LLM Outputs<\/title>\n<meta name=\"description\" content=\"Ricoh develops Safeguard Model to detect and block harmful LLM outputs, enhancing multi-layered AI safety for enterprise deployments.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/itbusinesstoday.com\/ja\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/\" \/>\n<meta property=\"og:locale\" content=\"ja_JP\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Ricoh Strengthens AI Safety with Guardrail for LLM Outputs\" \/>\n<meta property=\"og:description\" content=\"Ricoh develops Safeguard Model to detect and block harmful LLM outputs, enhancing multi-layered AI safety for enterprise deployments.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/itbusinesstoday.com\/ja\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/\" \/>\n<meta property=\"og:site_name\" content=\"itbusinesstoday\" \/>\n<meta property=\"article:published_time\" content=\"2025-12-26T11:24:49+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2025-12-29T13:19:13+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/12\/Ricoh-updates-guardrail-model-to-handle-harmful-LLM-output-detection-1.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1200\" \/>\n\t<meta property=\"og:image:height\" content=\"800\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"ITBT StaffWriter\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u57f7\u7b46\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"ITBT StaffWriter\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u63a8\u5b9a\u8aad\u307f\u53d6\u308a\u6642\u9593\" \/>\n\t<meta name=\"twitter:data2\" content=\"4\u5206\" \/>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u30ea\u30b3\u30fc\u3001LLM\u306e\u51fa\u529b\u306b\u5bfe\u3059\u308b\u300cGuardrail\u300d\u3092\u5c0e\u5165\u3057\u3001AI\u306e\u5b89\u5168\u6027\u3092\u5f37\u5316","description":"\u30ea\u30b3\u30fc\u306f\u3001\u6709\u5bb3\u306aLLM\u306e\u51fa\u529b\u3092\u691c\u77e5\u30fb\u906e\u65ad\u3059\u308b\u300cSafeguard Model\u300d\u3092\u958b\u767a\u3057\u3001\u4f01\u696d\u5411\u3051\u5c0e\u5165\u306b\u304a\u3051\u308b\u591a\u5c64\u7684\u306aAI\u306e\u5b89\u5168\u6027\u3092\u5f37\u5316\u3057\u307e\u3057\u305f\u3002.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/itbusinesstoday.com\/ja\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/","og_locale":"ja_JP","og_type":"article","og_title":"Ricoh Strengthens AI Safety with Guardrail for LLM Outputs","og_description":"Ricoh develops Safeguard Model to detect and block harmful LLM outputs, enhancing multi-layered AI safety for enterprise deployments.","og_url":"https:\/\/itbusinesstoday.com\/ja\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/","og_site_name":"itbusinesstoday","article_published_time":"2025-12-26T11:24:49+00:00","article_modified_time":"2025-12-29T13:19:13+00:00","og_image":[{"width":1200,"height":800,"url":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/12\/Ricoh-updates-guardrail-model-to-handle-harmful-LLM-output-detection-1.webp","type":"image\/webp"}],"author":"ITBT StaffWriter","twitter_card":"summary_large_image","twitter_misc":{"\u57f7\u7b46\u8005":"ITBT StaffWriter","\u63a8\u5b9a\u8aad\u307f\u53d6\u308a\u6642\u9593":"4\u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#article","isPartOf":{"@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/"},"author":{"name":"ITBT StaffWriter","@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#\/schema\/person\/94fb575e469aff25a4b938814d050f15"},"headline":"Ricoh Strengthens Generative AI Safety with New Guardrail Model for Harmful LLM Output Detection","datePublished":"2025-12-26T11:24:49+00:00","dateModified":"2025-12-29T13:19:13+00:00","mainEntityOfPage":{"@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/"},"wordCount":891,"publisher":{"@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#organization"},"image":{"@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#primaryimage"},"thumbnailUrl":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/12\/Ricoh-updates-guardrail-model-to-handle-harmful-LLM-output-detection-1.webp","keywords":["guardrail architecture","Guardrail Model","itbusinesstoday","LLM Output Detection","News","Safeguard Model"],"articleSection":["Cybersecurity","News Articles","Tech"],"inLanguage":"ja"},{"@type":"WebPage","@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/","url":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/","name":"\u30ea\u30b3\u30fc\u3001LLM\u306e\u51fa\u529b\u306b\u5bfe\u3059\u308b\u300cGuardrail\u300d\u3092\u5c0e\u5165\u3057\u3001AI\u306e\u5b89\u5168\u6027\u3092\u5f37\u5316","isPartOf":{"@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#website"},"primaryImageOfPage":{"@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#primaryimage"},"image":{"@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#primaryimage"},"thumbnailUrl":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/12\/Ricoh-updates-guardrail-model-to-handle-harmful-LLM-output-detection-1.webp","datePublished":"2025-12-26T11:24:49+00:00","dateModified":"2025-12-29T13:19:13+00:00","description":"\u30ea\u30b3\u30fc\u306f\u3001\u6709\u5bb3\u306aLLM\u306e\u51fa\u529b\u3092\u691c\u77e5\u30fb\u906e\u65ad\u3059\u308b\u300cSafeguard Model\u300d\u3092\u958b\u767a\u3057\u3001\u4f01\u696d\u5411\u3051\u5c0e\u5165\u306b\u304a\u3051\u308b\u591a\u5c64\u7684\u306aAI\u306e\u5b89\u5168\u6027\u3092\u5f37\u5316\u3057\u307e\u3057\u305f\u3002.","breadcrumb":{"@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#breadcrumb"},"inLanguage":"ja","potentialAction":[{"@type":"ReadAction","target":["https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/"]}]},{"@type":"ImageObject","inLanguage":"ja","@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#primaryimage","url":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/12\/Ricoh-updates-guardrail-model-to-handle-harmful-LLM-output-detection-1.webp","contentUrl":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/12\/Ricoh-updates-guardrail-model-to-handle-harmful-LLM-output-detection-1.webp","width":1200,"height":800,"caption":"Ricoh Strengthens Generative AI Safety with New Guardrail Model for Harmful LLM Output Detection"},{"@type":"BreadcrumbList","@id":"https:\/\/itbusinesstoday.com\/tech\/cybersecurity\/ricoh-strengthens-generative-ai-safety-with-new-guardrail-model-for-harmful-llm-output-detection\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/itbusinesstoday.com\/"},{"@type":"ListItem","position":2,"name":"Ricoh Strengthens Generative AI Safety with New Guardrail Model for Harmful LLM Output Detection"}]},{"@type":"WebSite","@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#website","url":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/","name":"\u30a4\u30c3\u30c8\u30d3\u30b8\u30cd\u30b9\u30c8\u30a5\u30c7\u30a4","description":"","publisher":{"@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"ja"},{"@type":"Organization","@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#organization","name":"\u30a4\u30c3\u30c8\u30d3\u30b8\u30cd\u30b9\u30c8\u30a5\u30c7\u30a4","url":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/","logo":{"@type":"ImageObject","inLanguage":"ja","@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#\/schema\/logo\/image\/","url":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Asset-1.png","contentUrl":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Asset-1.png","width":663,"height":275,"caption":"itbusinesstoday"},"image":{"@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/itbusinesstoday-68b209.ingress-baronn.ewp.live\/#\/schema\/person\/94fb575e469aff25a4b938814d050f15","name":"ITBT\u30b9\u30bf\u30c3\u30d5\u30e9\u30a4\u30bf\u30fc","image":{"@type":"ImageObject","inLanguage":"ja","@id":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Favicon-1.png90b3cb23d1d9c08733d969d29ea2d30d","url":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Favicon-1.png","contentUrl":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Favicon-1.png","caption":"ITBT StaffWriter"},"description":"ITBT\u30b9\u30bf\u30c3\u30d5\u30e9\u30a4\u30bf\u30fc\u306f\u3001IT Business Today\u306e\u793e\u5185\u5bc4\u7a3f\u8005\u3067\u3001\u30bf\u30a4\u30e0\u30ea\u30fc\u306a\u30cb\u30e5\u30fc\u30b9\u3068QuickBytes\u306e\u914d\u4fe1\u3092\u62c5\u5f53\u3057\u3066\u3044\u307e\u3059\u3002IT\u3001\u30cf\u30a4\u30c6\u30af\u3001\u4f01\u696d\u30bd\u30ea\u30e5\u30fc\u30b7\u30e7\u30f3\u306e\u6700\u65b0\u52d5\u5411\u3092\u30ab\u30d0\u30fc\u3057\u3001\u7c21\u6f54\u3067\u30a4\u30f3\u30d1\u30af\u30c8\u306e\u3042\u308b\u6700\u65b0\u60c5\u5831\u3092\u8aad\u8005\u306b\u63d0\u4f9b\u3057\u3066\u3044\u307e\u3059\u3002","sameAs":["https:\/\/itbusinesstoday.com\/"],"url":"https:\/\/itbusinesstoday.com\/ja\/author\/itbt-staffwriter\/"}]}},"authors":[{"term_id":572,"user_id":7,"is_guest":0,"slug":"itbt-staffwriter","display_name":"ITBT StaffWriter","avatar_url":{"url":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Favicon-1.png","url2x":"https:\/\/itbusinesstoday.com\/wp-content\/uploads\/2025\/01\/Favicon-1.png"},"author_category":"","first_name":"ITBT","last_name":"StaffWriter","user_url":"https:\/\/itbusinesstoday.com\/","job_title":"","description":"ITBT\u30b9\u30bf\u30c3\u30d5\u30e9\u30a4\u30bf\u30fc\u306f\u3001IT Business Today\u306e\u793e\u5185\u5bc4\u7a3f\u8005\u3067\u3001\u30bf\u30a4\u30e0\u30ea\u30fc\u306a\u30cb\u30e5\u30fc\u30b9\u3068QuickBytes\u306e\u914d\u4fe1\u3092\u62c5\u5f53\u3057\u3066\u3044\u307e\u3059\u3002IT\u3001\u30cf\u30a4\u30c6\u30af\u3001\u4f01\u696d\u30bd\u30ea\u30e5\u30fc\u30b7\u30e7\u30f3\u306e\u6700\u65b0\u52d5\u5411\u3092\u30ab\u30d0\u30fc\u3057\u3001\u7c21\u6f54\u3067\u30a4\u30f3\u30d1\u30af\u30c8\u306e\u3042\u308b\u6700\u65b0\u60c5\u5831\u3092\u8aad\u8005\u306b\u63d0\u4f9b\u3057\u3066\u3044\u307e\u3059\u3002"}],"_links":{"self":[{"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/posts\/25865","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/comments?post=25865"}],"version-history":[{"count":1,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/posts\/25865\/revisions"}],"predecessor-version":[{"id":25866,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/posts\/25865\/revisions\/25866"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/media\/25920"}],"wp:attachment":[{"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/media?parent=25865"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/categories?post=25865"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/tags?post=25865"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/itbusinesstoday.com\/ja\/wp-json\/wp\/v2\/ppma_author?post=25865"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}