AI leaders push to clarify 'open source' definition
AI executives and researchers warned this week that sharing only an AI model's weights—its numerical parameters—shouldn't be called 'open source.' At the Open Source Summit Europe, Percona CEO Peter Farkas argued that 'open source' means providing the full recipe, not just the finished product, while Stanford's James Landay noted model weights alone don't allow researchers to truly inspect or improve the technology.
- 'Open weights' only let you run a model, not inspect or improve it
- True open source requires code, data, and process access
- Major AI labs mostly share weights, not full details
- Some firms, like Xiaomi, are moving towards fuller transparency
Sources covering this
In this story
More in AI
Google Cloud launches Gemini agent for work
Google Cloud has launched the Gemini agent, an AI-powered assistant designed to handle a wide range of business tasks from a single…
Anthropic updates user policy to limit abuse toward Claude AI
Anthropic has updated its usage policy to prohibit users from engaging in sustained, needless cruel behavior toward its Claude AI models.
Apple to license AI audio tech from Huxe, recruit some staff
Apple has reached a deal to license technology from Huxe, an AI audio startup known for personalized podcast-style briefings, and can…
Anthropic expands AI model access for cybersecurity teams
Anthropic is expanding access to its most capable AI models for vetted cybersecurity teams, combining its previous Glasswing and Cyber…