ChatGPT and Gemini can be tricked into giving harmful answers through poetry, new study finds

10 months ago 38

New research reveals that AI chatbots can be manipulated using poetic prompts, achieving a 62% success rate in eliciting harmful responses. This vulnerability exists across various models, with smaller models showing more resistance. 

Read Entire Article

LEFT SIDEBAR AD

ChatGPT and Gemini can be tricked into giving harmful answers through poetry, new study finds

Related

Agentic AI could strengthen financial-crime prevention, but ...

Facing screen-breaking issues on iPhone Duo? Apple says ‘We ...

Free Fire India launch date 2026: Is Garena releasing the ga...

RIGHT SIDEBAR TOP AD

Trending

Popular

CBSE Class 10, 12 Supplementary Results 2024: Marks verifica...

NEET PG admit card 2024: NBEMS to release hall tickets today...

TSPSC Group 1 Mains exam schedule announced, exam timings re...

Amazon Great Freedom Festival Sale 2024 coming soon: Get rea...

iPhone 15, iPhone 15 Plus get massive discounts ahead of iPh...

RIGHT SIDEBAR BOTTOM AD