Politically Charged Prompts Crash DeepSeek-R1 Into Producing Insecure Codes

Politically charged prompts crash DeepSeek-R1 into producing insecure codes

DeepSeek-R1’s code output becomes insecure when political topics are included, revealing hidden censorship and serious risks for enterprise deployments.

DeepSeek took off as an AI superstar a year ago - but could it also be a major security risk? These experts think so

(Image credit: Getty Images) Experts find DeepSeek-R1 produces dangerously insecure code when political terms are included in promptsHalf of the politically sensitive prompts trigger DeepSeek-R1 to refuse to generate any codeHard-coded secrets and insecure input handling frequently appear under politically charged promptsWhen it released in January 2025, DeepSeek-R1, a Chinese large language model (LLM) caused a frenzy and has since been widely…

CrowdStrike test finds political triggers weaken DeepSeek-R1’s code security

An independent review by cybersecurity technology company CrowdStrike found that code generated by DeepSeek-R1, a large language model (LLM) from China-based AI startup DeepSeek, becomes significantly less secure when prompts contain topics considered politically sensitive to the Chinese Communist Party (CCP). In a technical report, Stefan Stein, manager for Data Science, Counter Adversary Operations at […] The post CrowdStrike test finds politi…

DeepSeek R1 Slim Explained: Inside the Quantum Compressed Uncensored LRM

A team at Multiverse Computing says it has built a smaller, faster variant of China’s DeepSeek R1 and removed its political guardrails. The company calls it DeepSeek R1 Slim and claims it is roughly half the size of the original while keeping reasoning quality comparable. The public pitch centers on an “uncensored, 55%-compressed” R1 made available through an API and the AWS Marketplace, with the savings credited to a quantum-inspired compressio…