%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
Research teams from the University of York and the University of Calgary conducted a large-scale analysis of posts related to AI programming on Reddit, finding that AI programming agents are frequently causing problems due to excessive permissions: overwriting files, deleting important data, and even generating malicious code. The research team used a crawler to collect 3,801 posts tagged with large language models between February 2023 and March 2026, from which they selected 446 posts about the safety of programming AI and analyzed more than 6,000 user comments below the posts.
Looking at the time distribution, posts about AI programming on Reddit reached a peak around the summer of 2025, with the highest number of questions in July 2025, which coincides closely with the widespread adoption of AI programming tools. As more developers hand over code generation, modification, and even execution tasks to AI agents, security issues have also emerged.
Cursor leads in accident reports, different tools face different challenges
According to the tools, the most reported issues were from Cursor, followed by Claude, Codex, Copilot, Windsurf, VSCode, Replit, and Cline, among other mainstream tools. Notably, different tools expose different types of issues: Cursor faced the most "runtime security issues" and "unauthorized data access," affecting production environments, meaning that AI agents might access sensitive data or cause damage to online systems without the user's knowledge; while Claude encountered more risks related to integration with third-party tools, focusing on security issues when connecting with external services.
This study reveals a deeper contradiction behind the popularity of AI programming assistants: the more powerful the tool, the higher its permissions, and the greater its potential for destruction. When AI agents are granted the ability to read and write code repositories, execute terminal commands, or operate production environments, any misunderstanding or failure in security boundaries could lead to losses far exceeding those caused by human coding errors. For developers, this is both a clear understanding of the limits of tool capabilities and a practical reminder about permission management and manual review mechanisms—AI programming assistants can improve efficiency, but "giving it all the keys" is clearly not a wise choice.