In other AI news, OpenAI's latest large language model, GPT-4.5, was rumored to have leaked online, but the leak turned out to be fake. The model is said to have multi-modal capabilities across language, audio, vision, video, and 3D. Meanwhile, Sports Illustrated's CEO was fired after the publication used AI-generated stories with fake bylines. Microsoft is reportedly planning to use nuclear power for its AI operations, and Pope Francis expressed concerns about an AI-fueled technocracy.
Key takeaways:
- The European Union's AI Act, a landmark artificial intelligence law, has been criticized for its major blindspots, particularly around law enforcement's use of AI technologies and the export of AI to other nations.
- The law allows for the use of live facial recognition technology by police in certain cases, a move that has been criticized by groups like Amnesty International, who argue that it could lead to expansive surveillance systems.
- The law does not ban the export of certain types of "high-risk" AI to other nations, meaning that while Europeans won't be subject to certain surveillance products, companies in the EU can sell those tools to other countries.
- Despite these criticisms, the AI Act's policy details are still being fine-tuned and there are still opportunities for the fine print to shift. The final version of the bill likely won't be finalized until sometime in January.