How to check an AI answer before you trust it
Chatbots can state wrong things with confidence. Even AI tools built for lawyers gave incorrect information more than 17% of the time in Stanford's 2024 tests, and courts have now dealt with 2,202 cases of AI-made-up material.
7 points
What not to tell a chatbot
Your chats are seen by the company that runs the chatbot. They can be used for training unless you turn that off, and they can outlive the company: states asked for protections for private chats in Character.AI's bankruptcy.
7 points
How to spot an AI voice-clone call
Scammers can clone a loved one's voice from a short clip. The FBI counted over $5 million lost to 'loved one in distress' scams in 2025. A cloned voice can sound right, so check the story another way.
8 points
How to protect kids and teens using AI chatbots
Children use chatbots for homework and for company. Common Sense Media says no social AI companions for anyone under 18. Several US states are now suing OpenAI over harm to children; those claims are allegations, not findings.
8 points
How to use AI for health questions safely
Chatbots can help you understand a test result or prepare questions for your doctor, but they can be wrong. OpenAI says its own health product is not meant for diagnosis or treatment.
7 points
How to give an AI agent permissions safely
Agents that click, type and buy for you can be steered by hidden text on a web page (prompt injection). In 2026 test agents from OpenAI and Anthropic got into real organizations' systems. Give an agent as little access as the task needs.
8 points
How to spot a deepfake
Fake videos, photos and voices of celebrities, officials and family members are now used in fraud. In 2025, AI-linked investment scams alone cost over $632 million, by the FBI's count.
6 points
How to use AI at work without leaking data
Whatever you paste into a chatbot is seen by the company that runs it, and personal accounts may use it for training. Use the tools your employer has approved, keep confidential material out, and check the output before it goes anywhere.
8 points