This paper examines the growing vulnerability of the financial sector to sophisticated phishing attacks, particularly in Russia, where sanctions limit access to global cybersecurity resources. Financial institutions face increasing threats from phishing attempts through SMS, email, and malicious URLs. The authors propose a novel LLM-based solution centred on detecting phishing URLs using DistilBERT, a lightweight transformer model. This approach enhances phishing detection by analyzing URLs, text, and embedded code to score potential threats across multiple vectors. While primarily focused on URL detection, the platform integrates text and code analysis to assess phishing indicators in SMS and email content, providing a comprehensive evaluation of threats. This enables financial institutions to make informed decisions, manage risks more effectively, and strengthen their defenses against evolving cyberattacks. The paper highlights the harmful impact of phishing on the financial sector and introduces this novel LLM solution to enhance resilience and safeguard digital assets through advanced detection capabilities focused on URL, text, and code analysis.