Technology news and discussion on TalkFuse.
Six Chinese AI firms accused of aggressively copying US frontier models
US national security agencies have publicly accused six Chinese AI companies of running industrial-scale campaigns to extract, or "distil", the capabilities of leading American AI models, potentially saving China billions in development costs. The NSA, CISA and FBI jointly named DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI, alleging they have targeted models including variants of Claude, GPT, Gemini and Grok since at least late 2024, likely with the Chinese government's awareness. The agencies say this activity threatens America's lead in the AI race and are calling for coordinated action across the industry to stop it.
The alleged tactics include mass-purchasing fraudulent accounts to run coordinated queries, sometimes numbering in the millions, and using prompt injection to trick models into revealing hidden chain-of-thought reasoning, with DeepSeek reportedly asking models to narrate their reasoning step by step. To counter this, the agencies recommend firms tighten identity verification, watch for suspicious subscription-to-usage patterns, and detect proxy-based access designed to dodge geographic restrictions. Most strikingly, they suggest US firms should covertly serve suspected malicious accounts a deliberately weaker, less capable model without notifying the user, a proposal the agencies concede would be technically difficult to implement.
In short
- US names six Chinese firms accused of copying frontier AI models at scale
- Alleged tactics include fake accounts, mass queries and reasoning-extraction prompts
- Agencies suggest secretly downgrading suspected Chinese users to weaker models
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
If the allegations are accurate, coordinated large-scale extraction could let firms reproduce costly capabilities without making a comparable investment in research, weakening incentives to develop advanced models and potentially eroding US security and economic advantages. Providers have a reasonable interest in protecting their systems, and stronger identity checks and abuse monitoring could help limit misuse while preserving legitimate access.
The case against
The accusations come from US agencies with a stake in maintaining America’s lead, so they warrant scrutiny and independent evidence before being treated as established fact. More broadly, learning from model outputs can resemble ordinary competition and reverse engineering; covertly downgrading service to suspected users risks mistaken targeting, unfair treatment and reduced trust, especially when suspicion is difficult to verify.
Explained simply
AI models are built by training software on large amounts of data so it can generate text, images or other outputs. Developers can also use “distillation”: they ask an existing model many questions and use its answers to help train a smaller or cheaper one. This can reduce the time and cost of building a competing system.
The companies named are Chinese developers DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI; the models they are alleged to have targeted were made by US firms including Anthropic, OpenAI, Google and xAI. The US agencies making the accusations are the NSA, CISA and FBI. The companies’ role and any government involvement are matters of allegation in the report.
The dispute matters because access to another company’s model can affect the cost and pace of AI development, and because governments see advanced AI as a strategic industry. It also raises questions about how model makers should protect their systems while allowing legitimate users to access them.