Prompt-Injection on Sameer Ankalgihttps://www.sameerankalgi.com/tags/prompt-injection/Recent content in Prompt-Injection on Sameer AnkalgiHugoen© {year} Sameer Ankalgi · ZürichTue, 18 Aug 2026 01:00:00 +0200Your prompt injection defenses may not composehttps://www.sameerankalgi.com/ai-safety/prompt-injection-defenses-may-not-compose/Tue, 18 Aug 2026 01:00:00 +0200https://www.sameerankalgi.com/ai-safety/prompt-injection-defenses-may-not-compose/I benchmarked stacked prompt injection defenses across four open-weight models. Two defenses that each work fine alone can block 100% of benign traffic when you chain them.