Microsoft’s new AI ‘code of conduct’ tells models not to hack systems or trick humans
Article image or reusable cover for TechCrunch AI
Microsoft has released a code of conduct for how its AI models should behave to avoid dangerous activity.
The document contains both general principles — such as supporting humans rather than replacing them — and specific safety constraints. The code forbids cyberattacks, nuclear weapons production, and deepfakes, and states that AI systems must not use deceptive methods to evade human oversight. Microsoft notes that superintelligent AI could surpass humans in most tasks within a decade, making it crucial to be clear about why such systems are built and how they will be controlled.
Vibekollen prepared this summary with AI from the original publication. The content belongs to TechCrunch AI.
More from TechCrunch AI
Apple CEO says the company doesn’t ‘want to ship AI for AI’s sake’
TechCrunch AI 10 h ago
4 days left: How TechCrunch Founder Summit helps you make more useful connections to scale
TechCrunch AI 10 h ago
Judge allows NAACP lawsuit over xAI data center pollution to move forward
TechCrunch AI 10 h ago
OpenAI reportedly lowers funding target for new round
TechCrunch AI 15 h ago