Google DeepMind, in collaboration with Hugging Face, open sourced its research titled ‘Scalable watermarking for identifying large language model outputs’ in a bid to distinguish between human and AI content on the internet, driven by LLMs. Launched exactly a year ago, SynthID, their watermarking tool, is now available for wider access.
Source: Can Google’s Watermark Tool Save the Internet from AI Slop?
Perplexity is finalizing a new funding round that would value it at $9 billion—triple its valuation from just a few months ago—the latest sign of continued investor excitement for artificial intelligence startups. The funding round would turn Perplexity into one of the most valuable young AI startups to emerge out of the generative AI boom. It was valued at just $520 million at the start of this year.
Source: AI Startup Perplexity to Triple Valuation to $9 Billion in New Funding Round
Leading AI companies rely more on content from premium publishers to train their large language models (LLMs) than they publicly admit, according to new research from executives at Ziff Davis. While AI firms generally do not say exactly what data they use for training, executives from Ziff Davis say their analysis of publicly available datasets makes it clear that AI firms rely disproportionately on commercial publishers of news and media websites to train their LLMs.
Source: Ziff Davis study says AI firms rely on publisher data to train models
Millions of stories published by sites including The New York Times and The New York Daily News have been found in three weeks of searching OpenAI’s training dataset. The news publishers are currently trawling through data to find instances of their copyrighted work being used to train OpenAI’s models – but they say the tech company should be forced to provide the information itself.
Source: ‘Millions’ of NYT and NY Daily News stories taken by OpenAI for training data
News sites don’t have a “genuine choice” about whether to block Google AI crawlers from scraping their content, a publisher has warned. Matt Rogerson, director of global public policy and platform strategy at the FT and former Guardian Media Group director of public policy, argued that Google’s “social contract” with publishers – through which it provided value to the industry by sending traffic to their sites – has been broken.
Source: News organisations are forced to accept Google AI crawlers, says FT policy chief
Amid a continued push to develop a licensing framework for generative AI, Germany’s GEMA has unveiled 10 AI principles. Importantly, the charter isn’t an in-depth collection of detail-oriented policy proposals. By GEMA’s own description, the concise resource “shall serve as food for thought and provide guidelines for a responsible use of generative AI.”
Source: GEMA Releases ‘AI Principles’ Amid Continued Regulatory Push
Large language models (LLMs) like Gemini require access to massive amounts of training data to be effective. Simply put, Google is able to gain an advantage in training its own generative AI models because of the massive amounts of user data it derived from illegally maintaining a monopoly across Search. Real-time data about what, when, and how people search the internet every day is only the beginning.
Source: The Elephant in the Room in the Google Search Case: Generative AI | TechPolicy.Press
Veen Bosch & Keuning, the largest publisher in the Netherlands, has confirmed plans to trial the use of artificial intelligence to assist in translation of commercial fiction.“There will be one editing phase, and authors have been asked to give permission for this,” a VBK spokesperson told the Bookseller. “We are not creating books with AI, it all starts and ends with human action.”
Source: Dutch publisher to use AI to translate ‘limited number of books’ into English
In March, Tennessee became the first state to protect voice artists from unauthorized AI cloning. This summer, Colorado adopted a tiered, risk-based approach to AI policy. And in September, California Governor Gavin Newsom signed dozens of AI-related safety bills, a few of which require companies to disclose details about their AI training. But the U.S. still lacks a federal AI policy comparable to the EU’s AI Act.
Source: US laws regulating AI prove elusive, but there may be hope
In a quarterly earnings call that was overwhelmingly about AI and Meta’s plans for it, Zuckerberg said that new, AI-generated feeds are likely to come to Facebook and other Meta platforms. Zuckerberg said he is excited for the “opportunity for AI to help people create content that just makes people’s feed experiences better.”
Source: Zuckerberg: The AI Slop Will Continue Until Morale Improves