IT之家 9 月 26 日消息,据 Wccftech 报道,一则关于“大语言模型破解付费软件”的帖子今日在 Reddit 上引发关注。
使用者 No_Ideal8394 声称,他利用无审查限制的 Qwen3.8-Flash-Next-Uncensored,成功破解了 Internet Download Manager(IDM)6.43 版本,使其绕过 IDM 试用限制,将软件变为完全激活版本,从而省去每年每台电脑 11.95 美元(IT之家注:现汇率约合 80.4 元人民币)的授权费用。在这一过程中,他只花了 5 欧元(现汇率约合 38.3 元人民币)去租用云 GPU,将难题交给模型处理。
发帖者表示,他向 AI 给出了约 25 万个 Token 的上下文信息,并让模型直接分析本机可执行文件,约 30 分钟后 AI 给出了可用的补丁思路。他还称,整个过程没有人工逐步指导,只需告诉 AI“我想要这个,去搞定它”。
依据其展示的截图信息,AI 直接分析本机可执行文件,并定位到了与授权验证相关的位置。模型只需修改其中一个字节,即可移除 IDM 注册检查,使软件表明为已激活状态。
发帖者承认,过去找到 IDM 密钥并不困难。但在 AI 时代,直接用 LLM 自主打补丁预计会快很多。该使用者称,自己原本期待不高,但结果令人意外。其表示难以想象前沿模型将来能做到什么。
这一事件引发了关于 AI 自动分析软件能力的讨论。过去,破解工具、注册机等通常需要人工开展逆向分析,而随着大语言模型结合代码分析能力的发展,部分研究者认为,AI 或将进一步降低软件分析与自动化修改的门槛。但这并不意味着所有软件都能被轻易破解,商业软件通常还会使用代码混淆、完整性校验、在线授权验证和硬件绑定等多种保护机制。
业内人士也指出,AI 可以辅助理解代码、定位难题或生成程序修改建议,与实际绕过商业软件授权之间存在显著区别。后者不仅涉及工艺难度,还涉及软件版权、许可协议和法律责任等难题。
本文来源:https://www.ithome.com/1/007/441.htm
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。
IT Home, September 26 news. According to Wccftech, a post about "large language model cracking paid software" attracted attention on Reddit today.
User No_Ideal8394 claimed that he used the uncensored Qwen3.8-Flash-Next-Uncensored to successfully crack Internet Download Manager (IDM) version 6.43, bypassing IDM's trial restrictions and turning the software into a fully activated version, thereby saving the license fee of $11.95 per computer per year (IT Home note: approximately RMB 80.4 at current exchange rate). In this process, he spent only 5 euros (approximately RMB 38.3 at current exchange rate) to rent a cloud GPU and handed the problem to the model.
The poster said he provided the AI with about 250,000 tokens of context and asked the model to directly analyze the local executable file. After about 30 minutes, the AI provided a feasible patch approach. He also said that the entire process required no step-by-step human guidance; he only needed to tell the AI, "I want this, go get it done."
Based on the screenshots he displayed, the AI directly analyzed the local executable file and located the part related to license verification. The model only needed to modify one byte to remove the IDM registration check, making the software indicate that it is activated.
The poster admitted that finding IDM keys wasn't difficult in the past. But in the AI era, using an LLM to autonomously patch is expected to be much faster. The user said he originally had low expectations, but the result was surprising. He said it's hard to imagine what frontier models will be able to do in the future.
This incident has sparked discussions about AI's ability to automatically analyze software. In the past, cracking tools, keygens, etc. usually required manual reverse analysis. With the development of large language models combined with code analysis capabilities, some researchers believe that AI may further lower the threshold for software analysis and automated modification. However, this does not mean all software can be easily cracked. Commercial software typically uses multiple protection mechanisms such as code obfuscation, integrity checks, online license verification, and hardware binding.
Industry insiders also pointed out that AI can assist in understanding code, locating problems, or generating program modification suggestions, which is significantly different from actually bypassing commercial software licenses. The latter involves not only technical difficulty but also issues such as software copyright, licensing agreements, and legal liability.

