Answer data protection questions where you already work

Yes, through frag.hugo. You connect Claude to your frag.hugo account and say in the chat what should be recorded. Claude creates it in the app. Whatever you submit for review reaches your data protection officer like any other entry; they approve it or come back to you with questions.

Yes, once ChatGPT is connected to frag.hugo. You list your software and processes, ChatGPT picks matching templates from more than 600 and creates the entries. If something is missing, it asks. Then you submit the record for review, and your data protection officer sees it.

Claude only reads and writes what your account may do in the app. Whatever is in the chat and whatever frag.hugo returns is processed by Anthropic as the provider of Claude, so use a plan with a data processing agreement for personal data, such as Claude Team. At frag.hugo every change is logged; we do not store your chats.

Yes. The Claude and ChatGPT apps accept voice on phone and computer. You might say: “The three new starters on Monday get the fire safety briefing, due in four weeks.” The assistant turns that into entries in frag.hugo. We do not listen in; our own app has no microphone.

Every change made through the assistant is listed in the app under “What your assistant changed”, with time, before and after. You can undo changes to entries there or in the chat for 90 days; sent emails cannot be recalled. Emails to people and deletions are shown as a preview first and only run after your yes.

No. The assistant takes the typing off your hands and records what you tell it. Whether an entry is right, whether a DPIA is needed and what belongs in the privacy policy is checked by your data protection officer, who remains your contact. The assistant works for them, not instead of them.

The ten public tools are free and need no account. With a booked frag.hugo area, administrators also get the tools for filling things in, at no extra charge. They can do what your plan can do in the app: with Hugo Learn you assign training, with Hugo DSB you maintain the record of processing.

Claude on the web, in the desktop app, on mobile and in Claude Code. ChatGPT and Microsoft 365 Copilot are in review at OpenAI and Microsoft; in ChatGPT you can already add frag.hugo as your own app in developer mode. Technically frag.hugo is an MCP server with OAuth sign-in.

No. We only see what the assistant passes to a tool, such as a search term or a domain. For the public tools we neither store nor log these inputs. The one exception is the website check: like every Hugo Check, the scanner logs the address it checks.

Yes. Whatever a tool returns becomes part of the conversation and is seen by the provider of your assistant, for example Anthropic, Microsoft or OpenAI. For personal data, use a plan with a data processing agreement; the DSGVO-Prüfstand (in German) shows which plans qualify.

No. For a data breach the assistant first shows a preview with the deadline and any missing details. Only after you explicitly confirm does frag.hugo create the record in your account. Nothing is sent to the supervisory authority automatically: you get a draft notification and report through the authority’s own channel.

No. It states the 72-hour deadline of Art. 33 GDPR and the competent authority. Whether a report is required is decided by your company and your data protection officer. The assistant is no substitute for legal advice in an individual case.

In the frag.hugo app, under Settings, you see every connected application with its date and can disconnect each one. A connection also ends by itself after 90 days without use.

No, to your account. The assistant may only do what your account may do in the app: if you cannot log data breaches there, you cannot do it through the assistant either.