1 Data use
Are your prompts, uploads, voice samples or generated outputs used to improve models? Can you opt out?
The terms matter because AI tools can receive your words, files, images, voice, calendar, inbox, code, business plans and habits. The risk changes with the job.
You do not need to become a lawyer. You do need to know the basic deal before a tool becomes part of your life.
Are your prompts, uploads, voice samples or generated outputs used to improve models? Can you opt out?
Can staff or contractors review your prompts or files for safety, support or model improvement?
Where is the company based, where may your data be processed, and which laws might apply?
Does the company sell to defence, intelligence, policing or national-security customers? Does that matter for your values?
Who runs it, who funds it, and does the business model reward careful service or endless data capture?
Look for lawsuits, safety resignations, copyright disputes, privacy complaints and misuse history.
Look for public-interest work: open research, science impact, accessibility, education, safety tooling, community benefit.
Can you delete data, cancel easily, export your work, turn off training and avoid lock-in?
This is a small habit. It is especially important before connecting email, Drive, calendars, voice samples, private photos or work files.
When you skim terms, look for these exact ideas. If you cannot find them, that is itself useful information.
Can your inputs or outputs train future models? Is opt-out easy? Are business plans treated differently?
Who owns generated output? Are there limits for commercial use, music, likeness, voice or brand work?
If the tool is wrong and you lose money, what does the provider actually promise? Usually, not much.