– Midjourney made a change in its terms of service related to IP disputes, indicating its confidence in winning court battles with creators over training data
– AI vendors are facing challenges with copyright issues and using copyrighted works to train generative AI models
– Other recent AI stories include EU authorities monitoring AI platforms for electoral shenanigans, OpenAI dismissing claims made by Elon Musk, and researchers using AI to predict COVID-19 variants and design molecules.
Midjourney, an AI startup specializing in image and video generators, recently updated its terms of service regarding IP disputes, suggesting an optimistic outlook on AI vendors prevailing in legal battles over training data. However, concerns have been raised about the use of copyrighted works in training data by vendors like Midjourney. While some vendors have established licensing agreements and opt-out options, others, like Midjourney, have used copyrighted material boldly without proactive measures. This risky strategy may pose a threat to Midjourney’s future if legal challenges arise.
Other recent AI developments include controversies surrounding ad creation on Instagram, EU authorities monitoring AI platforms ahead of elections, Google Deepmind’s AI-assisted gaming partner, and benchmarks used by many AI vendors which may have flawed metrics. Additionally, AI2 Incubator secured $200 million in funding, India debated government regulation for AI, and Anthropic introduced new models aiming to compete with OpenAI’s GPT-4.
Researchers are exploring AI applications in predicting COVID-19 variants, molecular design, and analysis of satellite imagery to measure methane emissions. Concerns have been raised about the impact of training large language models predominantly on English data, potentially affecting their ability to use and understand other languages effectively. Ethical considerations and regulation in AI research and development remain a topic of debate.