Sergey Sviridov

Title: Innovations of Sergey Sviridov in Cybersecurity

Introduction

Sergey Sviridov is an accomplished inventor based in Santa Clara, CA (US). He has made significant contributions to the field of cybersecurity, particularly in the detection of malicious instructions in large language models. His innovative approach addresses the growing concern of indirect prompt injection attacks, which pose a threat to the integrity of AI systems.

Latest Patents

Sergey holds a patent titled "Detection of indirect prompt injection attacks with malicious instructions detection models." This patent describes a malicious instructions detection model, referred to as a 'detector,' which intercepts augmented prompts intended for a large language model (LLM). The detector tokenizes and preprocesses sentences in the augmented prompts, obtaining confidence scores to identify potentially malicious instructions. If any confidence score exceeds a predetermined threshold, the detector blocks the prompt and generates an alert. Otherwise, it forwards the prompt to the intended LLM.

Career Highlights

Sergey Sviridov is currently employed at Palo Alto Networks, Inc., a leading cybersecurity company. His work focuses on enhancing the security measures of AI systems, ensuring that they are resilient against various forms of cyber threats. His expertise in this area has positioned him as a valuable asset to his organization.

Collaborations

Sergey collaborates with Chien-Hua Lu, a fellow innovator in the field. Together, they work on advancing technologies that protect against emerging cybersecurity threats.

Conclusion

Sergey Sviridov's contributions to cybersecurity through his innovative patent demonstrate his commitment to enhancing the safety of AI systems. His work at Palo Alto Networks, Inc. continues to influence the industry positively.