More from From ChatGPT to Gemini: how AI is rewriting the internet
Less than a week after its release to Gemini Advanced subscribers, the Google Gemini 2.5 Pro (experimental) model is now available to anyone who wants to try it, albeit with rate limits. Google says the new models now incorporate step-by-step “thinking” processes to become the best ones yet, and outpace the competition on LLM benchmarks.
Google says it’s also working on adding the new model to its Gemini mobile app, while Advanced subscribers still have access to a larger context window, and they can now use the new model in Gemini’s Canvas tool for live coding projects.
Molly White’s newsletter cogently outlines problems with how both AI companies and open access creators are approaching copyright, plus some possible solutions:
“The true threat from AI models training on open access material is not that more people may access knowledge thanks to new modalities. It’s that those models may stifle Wikipedia and other free knowledge repositories, benefiting from the labor, money, and care that goes into supporting them while also bleeding them dry.”
[citationneeded.news]
Google made the change to the feature, which adds app integrations to Gemini, in a beta version of the Google app for Android last week, later mentioning it in a Workspace weekly recap published Friday.
The recap adds that Gemini Apps are now powered by Gemini Flash 2.0, Google’s latest small on-device AI model, bringing “improved performance and better advanced reasoning capabilities with efficiency and speed.”
[workspaceupdates.googleblog.com]
Call center firm Teleperformance SE is rolling out an artificial intelligence system that “softens English-speaking Indian workers’ accents in real time,” aiming to “make them more understandable,” reports Bloomberg.
The company wouldn’t disclose which clients use the tech, Bloomberg writes. The article notes that Teleperformance’s call center and content moderation clients include Apple, Samsung Electronics, and TikTok.
[bloomberg.com]
Responses from 5,273 employed adults in the US show that 52 percent are worried about the use of AI in the workplace. People who reported using AI were more likely to say they believe it will affect future job opportunities, whether saying it would lead to fewer (42 percent) or more (15 percent), compared to 32 and 6 overall, respectively. The most common uses? Doing research or editing written content.
You can read the full breakdown right here, including questions and methodology.
That’s according to Google’s AI tool pricing tables, which it recently updated to include the early-access Veo 2 model.
As TechCrunch notes, that rate adds up to $30 per minute or $1,800 for an hour of AI-made video — far more than the $200-per-month subscription fee for OpenAI’s Sora.
[cloud.google.com]
Meta has announced Project Waterworth, a previously-rumored “multi-billion dollar, multi-year” plan to build an over 50,000-kilometer undersea cable to support AI infrastructure. It would be the world’s longest, designed to resist “damage from ship anchors and other hazards.”
Its planned route snakes from the east to west coasts of the US at up to 7,000 meters deep, connecting to Brazil, South Africa, India, and Australia along the way.
The Wall Street Journal reports that The Verge’s parent company, Vox Media, and other publishers like Conde Nast, Forbes Media, and Politico filed a copyright and trademark infringement lawsuit (pdf) against the enterprise AI company Cohere. They say evidence shows Cohere uses unlicensed copies of content to directly compete with publishers, and they list 4,000 specific examples of “verbatim regurgitations and substitutional summaries of news content.”
On the Decoder podcast, we recently discussed similar media lawsuits against AI firms and spoke to Cohere CEO Aidan Gomez last summer.
[wsj.com]
