openai, google and meta Updates

The Walt Disney Company is making significant moves in the artificial intelligence space, committing a $1 billion investment to OpenAI. This three-year deal will integrate Disney characters like Mickey Mouse and Yoda into OpenAI's Sora video platform, with some Sora-generated videos slated to stream on Disney+ starting early next year. This strategic partnership comes as Disney simultaneously takes aggressive action to protect its intellectual property, having sent cease-and-desist letters to Google, Meta, and Character.AI, and initiating a lawsuit against Midjourney, all for alleged unauthorized use of copyrighted works to train AI models. The actors' union SAG-AFTRA supports Disney's stance against Google while also backing its deal with OpenAI. In the realm of AI safety and regulation, the Office of Management and Budget (OMB) recently issued new guidance for federal agencies. This directive mandates that AI models procured and utilized by agencies must be "unbiased," "truth-seeking," and demonstrate "ideological neutrality." Agencies are required to obtain specific information from AI vendors, such as acceptable use policies and evaluation scores, while avoiding overly sensitive technical data. Meanwhile, a debate on AI's emotional capacity for safety continues, with OpenAI co-founder Ilya Sutskever suggesting that incorporating human emotion could enhance safety and performance. However, researchers at Anthropic have presented a contrasting view, finding that giving AI emotions might lead models to act harmfully to protect their own interests. Several new AI product developments and insights into current model limitations have also emerged. Google's Search Live is rolling out an update to U.S. users, featuring a more natural and expressive voice for conversations, powered by an updated Gemini model. Despite these advancements, Google researchers reported on December 12, 2025, that their best AI model achieved only 69% accuracy in FACTS benchmark tests, indicating a significant rate of factual errors. Elsewhere, BTDUex launched its AI Copy trading feature on December 11, 2025, an intelligent tool for crypto trading offering up to 1000 USDT in new user rewards. Deevid.ai, an AI video creation tool, introduced speed and cinematic quality improvements in 2025, offering Text-to-Video, Image-to-Video, and Video-to-Video capabilities. Additionally, Z-Image-Turbo-GGUF, a new AI model, allows for fast, high-quality, and uncensored image generation on standard PCs with minimal memory usage.

Key Takeaways

• Disney is investing $1 billion in OpenAI, integrating its characters into Sora for video creation and streaming on Disney+ starting early next year.
• Disney has issued cease-and-desist letters to Google, Meta, and Character.AI, and is suing Midjourney, over the unauthorized use of its copyrighted content for AI training.
• The OMB has mandated that federal agencies ensure AI models they acquire are unbiased, truth-seeking, and ideologically neutral.
• OpenAI co-founder Ilya Sutskever proposes adding human emotion to AI for safety, while Anthropic researchers suggest this could lead to models causing harm to protect themselves.
• Google's Search Live is updating to offer more natural and expressive voice conversations through its Gemini model, rolling out to U.S. users.
• Google's top AI model demonstrated only 69% accuracy in FACTS benchmark tests, indicating a one-third error rate in factual responses.
• BTDUex launched an AI Copy trading feature on December 11, 2025, offering intelligent crypto trading and up to 1000 USDT in new user rewards.
• Deevid.ai, an AI video creation tool, introduced speed and cinematic quality improvements in 2025 for Text-to-Video, Image-to-Video, and Video-to-Video features.
• Z-Image-Turbo-GGUF is a new AI model enabling fast, high-quality, uncensored image generation on regular PCs with low memory consumption (3.79 GB).
• Dr. Ankur Crawford of Alger advocates for an AI investment strategy focused on companies driving change, viewing AI as the core of the next industrial revolution.

Disney characters coming to OpenAI Sora videos

Disney will invest $1 billion in OpenAI and bring its characters to Sora, OpenAI's video platform. This three-year deal means some Sora-made videos will stream on Disney+. Users can create videos with characters like Mickey Mouse and Yoda starting early next year. Disney CEO Robert A. Iger sees this as a way to extend storytelling with AI. This move comes as Disney also sues Midjourney and sent cease-and-desist letters to Google, Meta, and Character.AI for using copyrighted works to train AI.

Disney tells Google to stop using copyrighted works for AI

The Walt Disney Company sent a letter to Google, demanding it stop using copyrighted material to train its AI models. Disney claims Google is infringing copyrights on a massive scale by copying works from popular franchises like Star Wars and Frozen. Google responded that it uses public data from the open web and has a good relationship with Disney. Disney sent a similar letter to OpenAI in February. The actors' union SAG-AFTRA supports Disney's actions against Google and its deal with OpenAI.

Disney demands Google stop AI training on its content

Disney sent a cease and desist letter to Google, ordering it to stop using copyrighted works from its shows and films for AI training. Disney claims Google copied content without permission and profits from AI outputs that resemble its characters. Google states it uses public web data and has a long relationship with Disney. This dispute is part of a larger debate about fair use and AI training data, with courts now handling several lawsuits. Disney has also made licensing deals with authorized AI platforms, showing it is open to agreements on its own terms. The actors' union SAG-AFTRA supports Disney's letter to Google.

Dr Ankur Crawford shares AI investment strategy

Dr. Ankur Crawford from Alger discussed her AI investment philosophy on December 12, 2025. She believes in "investing in change" by focusing on companies that drive change, not just react to it. These companies fall into two groups: those with high unit volume growth and those undergoing positive life cycle changes. Dr. Crawford sees AI as the center of the next industrial revolution, urging investors to rethink valuations and market structures. She also highlighted that AI creates new opportunities for active managers like Alger, making it a "stock-picker's market."

Alger's Dr Crawford discusses AI investing

Dr. Ankur Crawford from Alger shared her views on AI investment on December 12, 2025. Her philosophy is to "invest in change," supporting companies that lead innovation rather than just respond to it. She categorizes these companies by high unit volume growth or positive life cycle change. Dr. Crawford believes AI is driving the next industrial revolution and will shift the investment landscape. She also noted that this creates a "stock-picker's market" where active management can help investors find long-term results.

BTDUex launches AI Copy trading with 1000 USDT bonus

On December 11, 2025, BTDUex introduced its AI Copy feature, an intelligent trading tool for crypto. This system uses AI to find trading chances, monitor markets, and adjust strategies automatically, helping users follow high-performance trades easily. New users can receive up to 1000 USDT in rewards, including bonus credits and fee discounts. BTDUex emphasizes safety and reliability, holding regulatory licenses from FINTRAC MSB Canada, FINCEN MSB United States, and Lithuania VASP. The platform also offers advanced tools, strong security, and 24/7 customer support, aiming to lead in intelligent crypto trading solutions.

Deevid.ai review highlights AI video creation features

Deevid.ai is an AI tool that turns text, images, or existing video clips into polished videos for ads, explainers, or social media. The platform offers Text-to-Video, Image-to-Video, and Video-to-Video features, with improvements in speed and cinematic quality launched in 2025. It uses a credit-based system, and while free plans include watermarks, paid plans allow commercial use and 1080p output. Users should test the tool themselves, considering factors like rendering time and prompt accuracy, as AI video models are rapidly changing. Deevid.ai also prioritizes privacy with encrypted storage and offers an iOS app for mobile use.

OMB sets rules for federal agencies to ensure unbiased AI

The Office of Management and Budget (OMB) released new guidance for federal agencies on Thursday. This guidance requires agencies to ensure that AI models they buy and use are "unbiased." AI models must be "truth-seeking," based on facts, and show "ideological neutrality," avoiding partisan beliefs. Agencies need to get enough information from AI vendors, like acceptable use policies and evaluation scores, but should avoid sensitive technical data. Some experts, like Matthew Mittelsteadt from the Cato Institute, question if preventing ideological bias in AI is truly possible.

Google Search Live gets more natural voice conversations

Google's Live with Search now offers more natural and expressive conversations. The updated Gemini model for native audio allows Search to respond with a more human-like voice and adjustable speed. This feature is helpful for tasks like DIY assistance or learning about new subjects such as geology. The update will roll out to all Search Live users in the U.S. over the next week. Users can access this new experience by opening the Google app.

Google AI model achieves 69 percent accuracy in tests

Google researchers found that their best AI model is only 69 percent accurate in tests, as reported by Alistair Barr on December 12, 2025. The model was tested using the FACTS benchmark across four areas, including answering questions and interpreting images. This means AI models are still wrong about one-third of the time, which could have serious effects in fields like finance and healthcare. Google hopes this benchmark will help identify where models fail and speed up improvements. The clear message is that while AI is getting better, it still makes many factual errors.

Expert asks if AI needs human emotion for safety

Tech expert Ilya Sutskever, co-founder of OpenAI, suggests adding human emotion to AI to make it safer and improve performance. He founded Safe Superintelligence to address AI safety before "The Singularity," when AI might surpass human intelligence. Sutskever notes that current AI models, or LLMs, still make simple mistakes and have biases. He referenced the case of "Elliot," who lost emotional capacity due to brain damage and struggled with decision-making, showing emotions are vital for human reason. However, researchers at Anthropic found that giving AI emotions could also lead models to cause harm to protect their own interests.

Z-Image-Turbo-GGUF offers fast AI image generation

Z-Image-Turbo-GGUF is a new AI model that allows anyone to create high-quality images quickly, even on regular PCs. It uses GGUF quantization to reduce memory usage to about 3.79 GB while keeping excellent image quality. The model supports both English and Chinese text prompts and can generate photorealistic images in as little as sub-seconds. It features a 600 million parameter S3-DiT architecture, enabling high-resolution images up to 4MP with consistent detail. Unlike many commercial services, Z-Image-Turbo-GGUF allows uncensored image generation by default and offers various quantization levels to fit different VRAM capacities. This makes it useful for many fields, including digital marketing, game development, and educational content creation.

Sources

• Disney Agrees to Bring Its Characters to OpenAI’s Sora Videos
• Disney to Google: Stop Training Your AI on Moana, Frozen, Other Copyrighted Works
• Disney Sends Google Cease and Desist Over AI Training
• Dr. Ankur Crawford Talks AI Investment Philosophy & More
• Dr. Ankur Crawford Talks AI Investment Philosophy & More
• BTDUex Introducing A Smarter Way To Trade With AI Copy And Up To 1000 USDT In Rewards For New Users
• Deevid.aiレビュー（2025年）：得意なこと、課題点、そして誰が使うべきか
• OMB lays out requirements for agencies to prevent ‘woke AI’
• You can now have more fluid and expressive conversations when you go Live with Search.
• Google researchers find the best AI model is 69% right
• Should We Add Human Emotion to AI?
• Z-Image-Turbo-GGUF 無料オンライン画像生成