Goody-2

What Goody-2 is

Goody-2 is an AI assistant whose defining behaviour is refusal. It screens essentially every request as potentially harmful and declines, positioning itself as the ultimate in ethical compliance for legal, education and customer service contexts.

Read practically, it functions as a demonstration of what maximum safety-aligned behaviour looks like when taken to its limit, and is widely understood as satire of over-restricted models.

What you can do with it

• See how extreme refusal behaviour presents
• Discuss safety-utility trade-offs with a team
• Use it as a conversation starter on AI policy

Who it is for

• Teams discussing AI safety policy
• Curious users
• People designing their own model guardrails

What to watch out for

• It will not actually answer questions; that is the point
• Do not confuse its posture with a serious compliance solution
• The demonstration does not substitute for real AI governance work
• Its refusal reasoning is scripted humour, not a policy framework