Argh - My AI Agent deleted all my files

😱 “Argh - My AI Agent deleted all my files!!!!”

When was the last time one of your co-workers deleted an important file from your desktop?

I would hope it has been a long time ago and quite possibly never.

🖥️ Even at the time we used shared computers, home directories have been separated.

There was a notion of shared folders or network folders or whatever nomenclature the system you were using. So in order to provide access to files to coworkers, you would need to take the decision first to upload it to a shared folder.

🔒 Those systems came with permission management. Who can add, modify or delete what folders or files. A matter of configuration. Enforced by the system.

💻 Developers have long used specialized systems like git repositories, which provide all of this and versioning. Unintended deletion of files doesn’t mean a loss of that file. Just a version of the file. Quite possibly still there - just soft-deleted.

☁️ With the advent of Software as a Service systems, the notion of a file mostly got abstracted away. We don’t share files anymore, but maybe projects, plans, recipes, documents, or designs. What stayed is the notion of active sharing, permission management and versioning.

This is how we collaborate with co-workers, friends, and families.

🔄 That is why we don’t have the issue that co-workers delete our files from our desktops. They work on their own desktop. We share work items via centralized systems which allow us to recreate prior states.

So there is an isolation between co-workers, family members. They use their own account, workspace or whatever it is called.

🤖 Currently with the launch of Claude Cowork and similar services, I see a lot of people complaining, fearing, creating buzz around how dangerous AIs are in terms of possibly deleting your files, your work. Doing everything.

They are. Like your co-workers. Like your family members. Like all the humans you collaborate with.

𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗹𝗶𝗸𝗲 𝘆𝗼𝘂 𝗮𝗿𝗲.

❓ Anyone accidentally deleted an important file? Diploma thesis? Management Report? That is why we are using those permission-based and versioned systems.

I’m not going to anthropomorphize AI agents, but why exactly should we treat them with less care than we treat ourselves and our human co-workers? I’m blank here. There is no reason.

🔮 I get frequently asked by Customers, Partners and colleagues: What is the next big thing? What will keep us busy in 2026?

⚖️ 𝗪𝗲 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗯𝗮𝗹𝗮𝗻𝗰𝗲 𝗔𝗴𝗲𝗻𝗰𝘆 𝘃𝗲𝗿𝘀𝘂𝘀 𝗖𝗼𝗻𝘁𝗿𝗼𝗹. We want AI systems to be super easy to use, read our minds and just provide the answer we need. But we also need to make sure that nothing goes wrong. The more we control, the less agency we get. This is a balancing act.

✅ 𝗚𝗼𝗼𝗱 𝗻𝗲𝘄𝘀 𝗵𝗲𝗿𝗲: 𝗪𝗲 𝗸𝗻𝗼𝘄 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝗵𝗼𝘄 𝘁𝗼 𝗽𝗿𝗼𝘃𝗶𝗱𝗲 𝗰𝗼𝗻𝘁𝗿𝗼𝗹. We figured out how to build permission management and versioning. We also have already purpose-built frameworks and infrastructure at hand. Me working at AWS, Amazon S3 and Amazon Bedrock AgentCore comes to mind. But there are many more.

🛠️ Important is that we as Builders are using those building blocks to build AI Systems which allow us to nicely balance between agency and control.

👋 𝗪𝗲𝗹𝗰𝗼𝗺𝗲 𝘆𝗼𝘂𝗿 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁 𝗰𝗼𝘄𝗼𝗿𝗸𝗲𝗿 𝘁𝗼 𝘁𝗵𝗲 𝘁𝗲𝗮𝗺. 𝗠𝗮𝗸𝗲 𝘀𝘂𝗿𝗲 𝗶𝘁 𝗿𝗲𝗰𝗲𝗶𝘃𝗲𝘀 𝗶𝘁𝘀 𝗼𝘄𝗻 𝗹𝗮𝗽𝘁𝗼𝗽 - 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗮 𝘃𝗶𝗿𝘁𝘂𝗮𝗹 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝘄𝗶𝗹𝗹 𝗯𝗲 𝗴𝗼𝗼𝗱 𝗲𝗻𝗼𝘂𝗴𝗵, 𝗶𝘁𝘀 𝗼𝘄𝗻 𝗱𝗲𝘀𝗸, 𝗮𝗻𝗱 𝗮𝗰𝗰𝗲𝘀𝘀 𝘁𝗼 𝘀𝗵𝗮𝗿𝗲𝗱 𝗿𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀 𝘄𝗵𝗲𝗿𝗲 𝗶𝘁 𝗺𝗮𝗸𝗲𝘀 𝘀𝗲𝗻𝘀𝗲. 𝗟𝗶𝗸𝗲 𝘄𝗲 𝗱𝗼 𝗳𝗼𝗿 𝗵𝘂𝗺𝗮𝗻 𝗰𝗼-𝘄𝗼𝗿𝗸𝗲𝗿𝘀.

⚠️ We as users just need to make sure that we don’t do stupid things. Like we need to do to protect ourselves against us messing up and also how we establish a good platform to collaborate with other humans.

𝗟𝗼𝗼𝗸𝗶𝗻𝗴 𝗳𝗼𝗿𝘄𝗮𝗿𝗱 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗹𝗶𝗸𝗲 𝘁𝗵𝗶𝘀. 𝗛𝗮𝗽𝗽𝘆 𝘁𝗼 𝗱𝗶𝘀𝗰𝘂𝘀𝘀 𝗮𝗻𝗱 𝗰𝗼𝗻𝘀𝘂𝗹𝘁 𝗶𝗳 𝘆𝗼𝘂’𝗿𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗲𝗱. 𝗝𝘂𝘀𝘁 𝗽𝗶𝗻𝗴 𝗺𝗲!

---