Hyderabad: Hyderabad Police warn companies about rising prompt injection risks in AI systems, as Commissioner V.C. Sajjanar urged firms to adopt stronger security measures. He issued the warning on X and said the rapid spread of artificial intelligence had introduced a fresh cyber threat that organisations must not ignore.
He noted that firms now rely on AI chatbots for customer support, faster workflows and reduced costs. However, attackers are using malicious prompts to mislead AI models. As a result, they can extract internal documents, system details or confidential client records.
Hyderabad Police warn companies as AI prompt attacks rise
Sajjanar explained that prompt injection works by twisting the instructions given to AI. Consequently, these harmful prompts can confuse a model and reveal data that should remain hidden from users. He added that many companies link AI tools to sensitive systems such as CRM platforms, employee files and financial records. Because of this connection, a single deceptive prompt can expose restricted information.
He urged organisations to set up strong prompt guardrails. For example, he advised model-level safety training, strict prompt filters and tighter controls on APIs. He further said firms must conduct regular audits and restrict data permissions to essential users. Without such steps, he warned, operations may face disruption and valuable data may fall into criminal hands.
Moreover, Sajjanar said companies should adopt multi-layer security instead of depending on a single barrier. He stressed that only layered defences can reduce prompt injection threats and protect sensitive information from exploitation.
ఏఐకి ‘మాటల’ మాయ.. కంపెనీల డేటాకు ‘ప్రాంప్ట్’ గండం!
కృత్రిమ మేధ (AI) విస్తృతి పెరిగాక.. చిన్న స్టార్టప్ల నుంచి బడా మల్టీ నేషనల్ కంపెనీల వరకు అంతా ఇప్పుడు ‘ఏఐ చాట్బోట్’ల జపం చేస్తున్నాయి. కస్టమర్ల సందేహాలకు క్షణాల్లో సమాధానాలివ్వడం, పని వేగం పెంచడం, ఖర్చు తగ్గించడం వంటి… pic.twitter.com/rvdPuMGExe
— V.C. Sajjanar, IPS (@SajjanarVC) December 1, 2025