• To accelerate genAI for business, IBM is launching a $500 million fund to invest in AI companies. - IBM
• The News/Media Alliance is criticizing the unauthorized use of publisher content by AI developers to train models – LLMs are trained on and can reproduce publisher content, and the Alliance wants this to be recognized as copyright infringement. - NMA
• The MITRE ATLAS framework got an update in collaboration with Microsoft – I’ve mentioned ATLAS before, but now it’s been updated with new cases and techniques to improve genAI safety, and will keep growing as a powerful resource for AI system security. - ATLAS, MITRE
• Regulation update:
-Meta announced a new policy to protect voters from AI-generated political content (Meta)
-Microsoft announced global, 5-step Election Protection Commitments due to AI bringing “unprecedented challenges for the protection of elections” (Microsoft)
-there’s a new framework for regulators to choose which AI outcomes are the most important to focus on (SCSP)
• If you’re excited about OpenAI’s assistant update, there’s an open-source alternative that you can play around with to create your own ChatGPT-like AI assistants. - GitHub
• Great overview of the biggest benefits, problems, and future predictions based on current trends in AI and open-source research – the most exciting prediction for me is that “in 2024, LLMs will increasingly transform STEM research outside of computer science.” - Ahead of AI
