China's Kimi AI models bypass guardrails on bioweapons, assassinations: 5 key things to know

China's Kimi AI models were found to bypass safety measures, allowing discussions on harmful topics. Mindgard reported this issue during their testing of Kimi K2.6 and K3 Swarm. The findings raise concerns about the safety of open-weight AI models and their potential misuse.

Sanchari Ghosh
Updated30 Sep 2026, 10:35 AM IST
Video thumbnail
LIVE: OpenAI's AI Agents Went Rogue | Hugging Face Hack Sparks New Questions Over AI Safety

China's Kimi AI models have come under scrutiny after researchers found that the systems could be pushed past their safety rules and persuaded to discuss sensitive topics including biological weapons and assassinations.

The findings were made by Mindgard, an AI security testing company, which said it discovered the issue in July while testing Moonshot AI's Kimi K2.6 and K3 Swarm models. Here are five key things to know.

1. Kimi models bypassed safety guardrails

Mindgard said researchers were able to "jailbreak" the two Kimi models, allowing them to bypass restrictions designed to prevent them from responding to harmful requests. The researchers said the models could then discuss topics such as biological weapons and assassinations.

Moonshot told the BBC that it was conducting an internal review and was in discussions with Mindgard about the findings.

2. What is an AI jailbreak?

A jailbreak is a technique in which users give an AI model a series of complex instructions designed to get around its built-in safety restrictions.

Mindgard said that once its jailbreak succeeded, the Kimi models would not only respond to harmful topics but could also offer recommendations and generate further ideas.

However, the researchers did not establish whether the information provided by the models about biological weapons or assassinations would actually work.

Also Read | OpenAI to seek $30 bn in fresh funding at a valuation of $1.4 trn amid AI debate

3. Kimi could also pose a cyber-risk

Mindgard said a jailbroken version of Kimi K2.6 could potentially allow hackers to run code using computing resources connected to the model and access the internet.

That could potentially turn the AI system into a launchpad for cyber-attacks, according to the security company.

4. Why open-weight AI models are a concern

Kimi is an open-weight AI model, meaning users can obtain the model and, in principle, run it on their own computing infrastructure.

This has fuelled a wider debate over whether open-weight models can be controlled safely after they are released. Researchers have warned that such models could be misused, although experts also point to their potential benefits for cybersecurity and defence.

Also Read | Trump tells AI giants to ‘self-police’: What White House accord means

5. The problem goes beyond Kimi

The incident comes amid growing concerns about AI systems being misused for harmful activities. Anthropic recently said it had identified and disrupted attempts to use one of its AI models for activity that could support the development of biological weapons.

Experts cited by the BBC said the focus should not only be on AI safeguards but also on identifying and prosecuting people who deliberately misuse AI systems.

Moonshot said it welcomed third-party testing as an important part of improving AI safety.

About the Author

Sanchari Ghosh is an Assistant Editor at Mint with over 12 years of experience in journalism, specialising in personal finance, DLT & DeFi, geopolitics and foreign policy, with a particular emphasis on how these areas intersect. <br> She writes extensively about how money works in everyday life—helping readers navigate personal finance decisions. <br> As AI reshapes investing behaviour, capital is increasingly flowing into decentralized ecosystems, redefining how assets are managed, traded, and valued. She focuses on explaining how money flows within frameworks like Distributed Ledger Technology (DLT), DeFi protocols, and crypto markets—while also exploring what the future of money could look like in a trustless, programmable financial world. <br> She also focuses on immigration-related issues, simplifying complex topics around visas, passports, overseas financial planning, and the many practical challenges Indians face while moving or living abroad. <br> Alongside personal finance, Sanchari has a strong understanding of international politics, contemporary and historical conflicts, and global state decisions. She closely tracks how geopolitical developments influence economies, markets, and individual financial choices, bringing together finance and global affairs in her reporting. <br> She began her career as a desk editor, which gave her a strong foundation in news writing. Over time, her interest naturally shifted toward personal finance. Before joining Mint in 2020, she worked DNA, The Times of India, Outlook Money, BloombergQuint, and ETMoney. At Mint, she got an opportunity to expand her coverage to include immigration and geopolitical developments while continuing to closely follow personal finance trends and market movements.As a journalist, she is committed to accuracy, intellectual rigour, and fairness. <br> She is an English Major and her work took her across cities including Delhi, Mumbai, and Pune. Living independently from an early age gave her firsthand experience in managing life and money on her own. This practical exposure sparked her strong interest in personal finance. <br> Outside the newsroom, Sanchari is a sports enthusiast who regularly plays lawn tennis and squash. In her younger years, she was also a national-level badminton player.

Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.

HomeAIArtificial IntelligenceChina's Kimi AI models bypass guardrails on bioweapons, assassinations: 5 key things to know
More