Insights

RAG and Law: Why Reliable AI Cites Its Sources

3 min read · Editorial oversight: Avv. Federico Papa

AI adoption in the legal sector has reached a turning point: according to the 9th Report on the Legal Profession 2024 by CENSIS and Cassa Forense, 55.3% of Italian lawyers express a positive attitude or openness toward its use. However, simple text generation does not guarantee the accuracy required for a legal document or professional legal opinion.

In brief

The 9th CENSIS and Cassa Forense Report (2024) indicates that 55.3% of Italian lawyers are open to using AI. RAG (Retrieval-Augmented Generation) technology mitigates the risk of hallucinations by consulting authoritative sources prior to generating text. The negligent use of non-existent citations exposes lawyers to liability under Art. 96 c.p.c. and breaches the duty of diligence. EU Regulation 2024/1689 (AI Act) imposes transparency obligations under Art. 50 effective from 2026. Applicable legislation prohibits the complete delegation of legal drafting, requiring human oversight as emphasized by the National Bar Council and Art. 27 of the Code of Conduct for Lawyers.

1.
Beyond AI Memory: RAG Technology
Generalist AI models such as ChatGPT or Claude rely solely on their training memory, which can lead to hallucinations. RAG (Retrieval-Augmented Generation) technology allows the system to consult specific documents before responding. This approach extracts relevant passages from trusted sources to generate answers based on real data rather than statistical probabilities.
2.
The edit.legal Standard: Post-Generation Verification
Simple document retrieval is not sufficient for professional compliance and security. The edit.legal standard includes a citation verification phase that compares the generated output against over one million official documents, including statutory law and Supreme Court precedents. This cross-check ensures that every legal reference cited is authentic and accurate according to official databases.
3.
The Legal Risk of Hallucinations
Relying on unverified systems can lead to severe financial and ethical consequences. International cases of sanctions for frivolous litigation are well documented, as is the risk in Italy of condamnations for bad-faith litigation pursuant to Art. 96 of the Italian Code of Civil Procedure (c.p.c.). The negligent use of non-existent AI-generated citations exposes the lawyer to direct liability for breach of the duty of diligence.
Apply this research directly with edit.legal
Legal research and drafting with citations checked against official databases. Try edit.legal for free, no credit card.
Try edit.legal
4.
AI Act and New Transparency Obligations
While rules on general-purpose AI (GPAI) models and governance will apply from August 2, 2025, transparency obligations (Art. 50) under EU Regulation 2024/1689 (AI Act) will enter into force on August 2, 2026. Entities using AI systems must clearly mark synthetic content and inform users. Furthermore, AI systems intended for use by judicial authorities are classified as high-risk.
5.
The Human Contribution in Legal Assistance
Current legislation establishes that AI can only be used in a supporting capacity; the complete delegation of court document drafting to automated systems is strictly prohibited. Human oversight must remain central, a principle reaffirmed by the Italian National Bar Council regarding professional liability and the essential supervisory role of legal counsel.
6.
Operational Tools for the Modern Lawyer
Qualitative analysis of thousands of real legal queries demonstrates that efficiency stems from seamless integration. edit.legal offers a native editor with variable fields and automated Legal Basis generation, alongside a Microsoft Word add-in and an MCP server for Claude or Cursor. These tools enable practitioners to work across 21 practice areas using specialized multi-agent workflows tailored to the specific case.
7.
Data Security and European Infrastructure
Protecting sensitive client data is guaranteed by an entirely European infrastructure fully compliant with the GDPR. Unlike many foreign services, edit.legal offers transparent pricing and a free trial requiring no credit card, allowing legal professionals to test the accuracy of source verification without mandatory sales demos or hidden fees.

Editorial oversight: Avv. Federico Papa·ICAMContent drafted with AI support and subject to editorial source checks. Despite these controls, inaccuracies may remain: reports and rectification requests are welcome. Report a correction

Frequently asked questions

What is the difference between ChatGPT and a legal RAG system?

ChatGPT relies solely on its internal memory, whereas a legal RAG system reads relevant legal documents before generating an answer. edit.legal goes a step further by verifying every citation against official databases, such as the Official Gazette, to eliminate the risk of fabricated citations.

Can AI be sanctioned if it makes a citation error?

Sanctions apply to the lawyer, not the software, for breach of the duty of diligence and professional competence. Courts have already begun issuing sanctions for bad-faith litigation in cases where submissions relied on non-existent precedents generated by AI.

Do I need to inform my client if I use AI?

Yes, based on an expansive interpretation of the duty to inform (Art. 27 of the Italian Code of Conduct for Lawyers). Although the rules do not explicitly mention AI, the obligation to inform the client about how the mandate is performed advises transparency regarding the use of automated tools, which must always remain under human supervision.