Skip to main content

AI Assistant Privacy Grades

2 companies analysed · Sorted by privacy score

People treat chat assistants like confidants—pasting medical worries, legal questions, HR drafts, and family details—yet the default is often retention, safety review, and model improvement unless you change settings. The critical divides in this category are training on user content, retention windows, enterprise vs consumer terms, and whether “delete” removes data from training corpora or only from your UI. Stronger players minimise retention, separate consumer from workspace data, and document subprocessors and cross-border transfers clearly; weaker ones bundle broad licensing, long inference windows, and fuzzy human-access exceptions. Before adopting a tool for sensitive work, read how prompts, files, and plug-in data flow to third parties and whether opt-outs apply retroactively. Policies change quickly here, so we re-read when vendors ship major updates. Grades reflect the same methodology as the rest of the site—see the About page for how each pillar is scored.

#CompanyGradeScoreIn plain English
1
Anthropic logoAnthropic
B72/100Anthropic collects your conversations and account data, uses them to train AI models unless you opt …
2
OpenAI (ChatGPT) logoOpenAI (ChatGPT)
D42/100OpenAI collects account data, all prompts and responses, file uploads, voice inputs, and a separate …
How we grade·Each company is scored 0–100 across four pillars: data collection, third-party sharing, user controls, and policy promises. The overall grade maps to the score band. → Read the full methodology

More categories