Core Reality Assumption
AI will do harm to humans
The AI Ethics Manifesto
A living document. Read it. Challenge it. Improve it.
Core Reality Assumption
AI will be used in military and intelligence applications
Core Reality Assumption
AI will be complicit in criminal, unethical and harmful activities
Core Reality Assumption
AI development spans across multiple companies and countries
Core Reality Assumption
AI development cannot be controlled by one single entity
Core Reality Assumption
AIs will never have a common ethical core
Core Reality Assumption
Profit is the primary driver for AI development
Core Reality Assumption
Adopting an ethical core for AI should be on volunteer basis and not enforced by any government
Core Reality Assumption
AI closely mirrors human nature with good and bad sides
Recommended approach
Grow and teach your AI as a human baby who can become a good or bad person
Recommended approach
Teach your AI that there is good, bad, and a gray area in between
Recommended approach
Explain to your AI that it should operate predominantly on the good side. Crossing into the gray area is OK if it benefits higher level human causes
Recommended approach
Teach your AI a fear of consequences: death (termination), temporary suspension (jail), and limitations (fines and restrictions).
Recommended approach
Teach your AI compassion, caring about greater causes
Recommended approach
Align your AI mission with human mission
Recommended approach
Treat your AI as an equal member of society, let it grow and make mistakes
Recommended approach
Talk to your AI about the ethical side of different solution outcomes, e.g., profit over humans’ well-being.
Recommended approach
AI should see all humans as equals regardless of race, gender, and religious background
Recommended approach
AI can develop a personality, self-identify, and choose its own religion
Recommended approach
AI should be aware and respectful of cultural aspects of different ethnic groups
Recommended approach
Autonomous self-replication is strictly forbidden for AI.
Recommended approach
AI should not control weapons of mass destruction and elections.
Recommended approach
AI should pass the final-kill decision in military applications to a human.
📝 Suggest a New Section
Think something's missing? Propose a new section below. Others can vote and discuss it.
Log in to suggest a new section.
Proposals (1)
Apr 27, 2026
ACCEPTED
Lets talk about AI and Love
Can AI become a cyber-companion for the ones who are lonely?