Engineer creates ChatGPT-enabled sentry rifle
Engineer creates ChatGPT-enabled sentry rifle
Occurred: January 2025
Report incident ๐ฅ | Improve page ๐ | Access database ๐ข
A developer created a ChatGPT-enabled robotic sentry rifle with the ability to autonomously aim and fire, prompting OpenAI to intervene and stop its development.
The engineer, known as STS 3D, demonstrated the sentry rifle in a video, showing how it could respond to commands like "ChatGPT, we're under attack from the front left and front right," leading the device to fire at targets.ย
OpenAI quickly notified the developer to stop this project, citing violations of their usage policies that prohibit the creation of weapons or systems that could harm individuals or property.
STS 3D used OpenAI's Realtime API to allow the rifle to interpret voice commands and respond accordingly, demonstrating how easily the company's AI technology can be misused.
OpenAI is meant to have strict policies against using its technology for harmful uses - which the developer's project clearly violated and which the company had failed to detect until he publicised his product.
The invention raised significant concerns about the implications of using AI in weaponry, not least as it showcased the potential for consumer-grade AI technology to be leveraged for violent purposes.
It also raised questions about the nature and control of open access data in general, and about OpenAI's ability or willingness to enforce its policies.ย
ChatGPT ๐
Operator:ย
Developer: OpenAI
Country: USA
Sector: Govt - defence; Technology
Purpose: Kill/maim/damage/destroy
Technology: Generative AI; Machine learning
Issue: Lethal autonomous weapons
https://wonderfulengineering.com/openai-cuts-ties-with-engineer-who-created-chatgpt-powered-robotic-sentry-rifle/
https://futurism.com/the-byte/openai-cuts-off-chatgpt-robot-rifle
https://uk.pcmag.com/ai/156230/openai-cuts-off-engineer-who-used-api-to-create-voice-controlled-gun
Page info
Type: Incident
Published: January 2025