What are the main risks?
- It can be wrong. Generative AI predicts plausible text, so it can invent facts, figures or citations and sound sure about them.
- Your prompts go somewhere. What you type is processed by the provider, and depending on the service it may be stored or used to improve models.
- It can be steered. Text on a web page or in a document can contain instructions aimed at the AI.
- Actions are harder to undo than words. An AI that can send, book or buy needs more care than one that only answers.
What habits make it safe?
- Leave out passwords, card numbers, ID numbers and health or financial records. More in is it safe to paste personal info into AI.
- Check facts that matter against the original page or a trusted source.
- Look for a setting or mode that keeps chats out of history and training.
- Review anything before it is sent, posted or paid for.
- Use AI from known companies, downloaded from their official sites.
Is every generative AI tool equally safe?
No. Tools differ in what they store, who they share data with and how much they are allowed to do. Read the privacy policy, check whether chats are used for ads or training, and prefer tools that ask before taking an action. Doing that once per tool covers most of the risk.
How does Ace handle these risks?
Ace is a browser with an assistant that reads the page you ask about. Your conversations are not used for targeted ads, and Incognito Chats are never saved and never used for memory or model training. Ace acts on the page in front of you and does not buy or book on other sites unattended. Its privacy policy notes that data may be shared with cloud and AI providers and retained for up to 24 months after you stop using Ace, which is the kind of line worth reading in any tool. The rest is on the safety page.