A group of independent EU publishers has filed a complaint to the European Commission over Google's AI Overviews feature. They claim Google uses their content to generate AI-written summaries without permission, reducing web traffic and harming revenue. According to the complaint, publishers cannot opt out of the AI feature unless they also give up visibility in Google Search. The group is supported by Foxglove Legal and the Movement for an Open Web, and is requesting interim measures to stop what they describe as ongoing, irreparable damage. Google's AI Overviews pull information from websites to create short answers directly in search results without giving back traffic or compensation to the original publishers.
OpenAI is set to rent an additional 4.5 gigawatts of computing power from Oracle's US data centers for its Stargate AI project, according to Bloomberg. That amount of energy is roughly equivalent to the electricity use of several million households. To meet the demand, Oracle plans to build new data centers across multiple states, including Texas, Michigan, and Wyoming. The existing Stargate facility in Abilene, Texas, is expected to expand from 1.2 to 2 gigawatts. The deal is part of a broader cloud contract with Oracle valued at $30 billion per year.
Black Forest Labs has released FLUX.1 context [dev], a 12-billion parameter AI image model, for free non-commercial use on HuggingFace. It runs on standard hardware and supports ComfyUI, HuggingFace Diffusers, and TensorRT. According to Black Forest Labs and Artificial Analysis, it performs better than models like Gemini-Flash Image in most benchmarks. Optimized versions are available for Nvidia's new Blackwell architecture. For commercial use, licenses start at $999 per month via Black Forest Labs' license portal.
FLUX.1 context [dev] scores higher than many competing models in all six categories. | Image: Black Forest Labs
OpenAI is ramping up its enterprise AI consulting business, charging at least $10 million per client, according to The Information. The company's engineers work directly with organizations to adapt models like GPT-4o to their specific data and build custom applications, including chatbots. The push puts OpenAI in direct competition with established players like Palantir and Accenture. The team handling these projects is known internally as "Forward Deployed Engineers" (FDE).
Services go beyond model customization. OpenAI also offers data labeling, where experts review and correct AI-generated answers. Insiders say OpenAI is considering outsourcing this work to specialists like Snorkel AI and Surge AI. Its customer list includes the US Department of Defense and Southeast Asian tech company Grab.
![Bar charts show FLUX.1 Context [dev] with peak values for character preservation, editing accuracy, style transfer, and text input compared to other image AI models.](https://the-decoder.com/wp-content/uploads/2025/07/BFL_benchmark_results.png)
