#ai-training-data

#ai-training-data

Every search, purchase, loyalty swipe, location ping and scroll feeds systems that now shape pricing, product decisions, hiring and marketing strategies. Most founders understand this in theory, but few grasp the practical consequence: whether they intend to or not, they and their customers are already casting votes with their data. And those votes? They're usually cast passively, on someone else's terms.

The stock prices of some major video game companies, including Take-Two Interactive, Roblox, and Unity, had notable declines on Friday, just a day after Google announced its Project Genie tool that lets users prompt AI to generate interactive experiences, Reuters reports. Take-Two's stock price closed at $220.30 (down 7.93 percent from yesterday), Roblox's closed at $65.76 (down 13.17 percent), and Unity's closed at $29.10 (down 24.22 percent).

those options range from "option 0", simply doing nothing and leaving UK copyright legislation in its currently uncertain state when it comes to the use of copyright materials to train AI models, through to options which would either require specific consent from rights holders in all cases ("option 1") or allow consent to be assumed by AI developers unless a rights holder objects, subject to developers being transparent about what materials have been used in training ("option 3").

A proposed class-action lawsuit filed on behalf of Elizabeth Lyon, an author from Oregon, claims that Adobe used pirated versions of numerous books-including her own-to train the company's SlimLM program. Adobe describes SlimLM as a small language model series that can be "optimized for document assistance tasks on mobile devices." It states that SlimLM was pre-trained on SlimPajama-627B, a "deduplicated, multi-corpora, open-source dataset" released by Cerebras in June of 2023.

I contacted Google for comment, and a spokesperson sent me the following statement: "These reports are misleading - we have not changed anyone's settings. Gmail Smart Features have existed for many years, and we do not use your Gmail content for training our Gemini AI model. Lastly, we are always transparent and clear if we make changes to our terms of service and policies."

"The great patron of the internet for the last 27 years was Google. The great villain of the internet today is also Google," Prince said. He claimed that in the past, for every two pages that Google crawled to inform its search engine, it would, on average, send one visitor to those sites-traffic that publishers can monetise with advertising.

Wikimedia has called on AI companies to take responsibility for using Wikipedia content in their language models. This can be achieved by stopping scraping and using the paid API instead. In a blog post, the organization states that artificial intelligence cannot exist without the human knowledge collected and maintained on platforms such as Wikipedia. To maintain that balance, Wikimedia asks developers of generative AI to clearly cite their sources and contribute to the continued existence of the open knowledge project via the paid Wikimedia Enterprise platform.

Since 2001, Wikipedia has been the backbone of knowledge on the internet. Hosted by the Wikimedia Foundation, it remains the only top website in the world run by a nonprofit. Unlike newer projects, Wikipedia's strengths are clear: it has transparent policies, rigorous volunteer oversight, and a strong culture of continuous improvement. Wikipedia is an encyclopedia, written to inform billions of readers without promoting a particular point of view.

Canva has built its own foundational AI model that generates layered designs users can edit more easily. It's one of several generative AI-related features Canva announced Thursday, alongside expanded access to its AI assistant and content generation capabilities across its app. To date, Canva has partnered with a variety of AI model providers for content generation - Black Forest Labs, Google, and OpenAI among them - and it acquired Leonardo AI last year.

According to Uber, beginning later this year, drivers and couriers who opt into the program can complete "digital tasks" within Uber's Driver app. These tasks can include submitting a video of themselves speaking in their native language, uploading pictures of specific everyday items, or presenting documents written in a different language. After tasks are completed, the earnings will be in the users' balance within 24 hours. Compensation depends on the time commitment to complete tasks and their complexity.

The new change, which Cloudflare calls its Content Signals Policy, happened after publishers and other companies that depend on web traffic have cried foul over Google's AI Overviews and similar AI answer engines, saying they are sharply cutting those companies' path to revenue because they don't send traffic back to the source of the information. There have been lawsuits, efforts to kick-start new marketplaces to ensure compensation, and more-

"We've already run out of data," Neema Raphael, Goldman Sachs' chief data officer and head of data engineering, said on the bank's "Exchanges" podcast published on Tuesday.

For decades, publishers large and small have created the news, culture, entertainment, and educational resources that shape how society consumes information. Yet in recent years, the rise of artificial intelligence has added a new twist to the ongoing struggle for sustainable publishing. AI companies are building tools capable of generating responses, summaries, and insights trained on vast amounts of web content. The problem? Many publishers see little to no compensation for their role in shaping the data that fuels these systems.

When Is LinkedIn Going to Start Using My Data to Train Its AI Models? LinkedIn has announced that it will start using some of its users' data to train its AI models starting on November 3rd, 2025. Users from the EU, EEA, Switzerland, Canada, and Hong Kong will all be affected. At this stage, US users will not be affected, but this could soon change.

Media companies have filed so many lawsuits against AI companies over the past two years that the act has become routine. When I report on these in The Media Copilot newsletter, they're often digest items, adding to the pile of publishers who want fair compensation for the content AI labs have ingested to create large language models (LLMs). There are so many that elaborate infographics are required to keep track of them all.

When Google became the dominant search engine around 2004, not everyone was happy. Everyone from book publishers to music studios blasted the company for helping itself to copyrighted content without paying. The search giant eventually smoothed things over but now, twenty years later, Google has become the media industry's villain all over again-this time for gobbling that same content to train its AI tools.