OpenAI is quietly building a social network to gather real-time data, possibly challenging X and Meta. Kling AI is pushing creative AI forward with major updates to video and image models. Meanwhile, UC San Diego’s Hao AI Lab is testing AI’s detective skills in the Ace Attorney challenge, revealing the gap in logic and decision-making. Finally, Google has ramped up its fight against ad fraud, suspending millions of accounts with the help of large language models.

• OpenAI develops a social network to supercharge model training
• Kling AI unveils 2.0 models for cinematic video and image editing
• AI Plays Detective in Phoenix Wright: Ace Attorney challenge
• Google suspends 39M ad accounts using AI to prevent fraud

Special highlight from out Network

It’s Mode Mobile, named 2023’s fastest-growing software company by Deloitte.¹

With the EarnPhone and EarnOS, Mode has:

They’ve partnered with Walmart and Best Buy, and the momentum is still building.

Just like Uber turned cars into income tools, Mode is helping people turn smartphones into passive income.

Now, you have a chance to invest before their potential IPO at just $0.26 per share.

They’ve just been granted the Nasdaq ticker $MODE², and the share price is changing in two weeks.

_Disclaimers

_{1 Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur.}

_{2 The rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period.}

_{3 A minimum investment of $1,950 is required to receive bonus shares. 100% bonus shares are offered on investments of $9,950+.}

OpenAI is quietly prototyping a social network, according to multiple sources, potentially setting up a collision course with X and Meta. The internal project, which features ChatGPT’s image generation in a social feed, could give OpenAI access to real-time user data to supercharge model training while intensifying Sam Altman’s rivalry with Elon Musk.

What we know so far:
• An internal prototype features a social feed showcasing AI-generated images from ChatGPT
• CEO Sam Altman is privately soliciting feedback on the app from outsiders, sources say
• Unclear whether it will be a standalone product or built into ChatGPT, now the most downloaded app worldwide
• OpenAI did not comment, but Altman has previously teased a social app in response to Meta’s moves
• Musk recently offered $97.4B to buy OpenAI; Altman declined, joking he’d rather buy Twitter for $9.74B

A social platform could give OpenAI something it has long lacked; fresh, user-driven data in real time. With ChatGPT already topping app store charts, a built-in social layer could become a powerful feedback loop to train smarter models while drawing users away from X and Instagram

Chinese AI startup Kling AI has launched significant upgrades to its creative suite, unveiling KLING 2.0 Master for video and KOLORS 2.0 for images. These new models bring enhanced prompt adherence, realistic outputs, and cutting-edge editing capabilities.

What’s new:
• KLING 2.0 Master now handles sequential actions and expressions, producing cinematic videos with natural motion and speed.
• KOLORS 2.0 generates images in over 60 styles, ensuring realistic depth, color accuracy, and precise subject positioning.
• The model includes new editing features, such as inpainting to modify images and restyle to change their look.
• The 1.6 video model is updated with a multi-elements editor, allowing users to add, swap, or delete video elements using text input.

Chinese AI startups are quickly closing the gap with Western competitors like Veo and Sora. Kling AI’s advancements in video generation and editing are positioning it as a strong contender in the creative AI space. Early reactions suggest KLING 2.0 is narrowing the quality gap, and its multimodal editing features could redefine how creators interact with AI.

Researchers at UC San Diego's Hao AI Lab tested leading AI models on their ability to play Phoenix Wright: Ace Attorney, a video game where players investigate crime scenes and solve complex legal cases. The models were challenged to identify evidence, expose contradictions, and crack the case.

How does it work:
• Top models, including GPT-4.1 and Gemini 2.5 Pro, were tasked with playing Phoenix Wright, who uncovers evidence and cross-examines witnesses.
• o1 and Gemini 2.5 Pro excelled, presenting 26 and 20 correct pieces of evidence, respectively, reaching level 4—but neither solved the case.
• Other models, including the new GPT-4.1, struggled, matching only 6 correct pieces of evidence, tied with the older Claude 3.5 Sonnet.
• Llama-4 Maverick and others fared poorly, presenting fewer than 10 correct pieces of evidence.

Ace Attorney tests multiple AI capabilities, from visual understanding to decision-making. The AI models' performance highlights the gap in AI's ability to handle real-world logic, cross-referencing, and long-term reasoning. As AI improves, games like this will push models to evolve and tackle more complex interactive challenges.

Google recently announced a massive crackdown on ad fraud, suspending over 39 million advertiser accounts in 2024 whichmore than three times the previous year's figure. The company leveraged large language models (LLMs) to identify suspicious activity, such as business impersonation and fraudulent payment details, helping it stop fraud before ads even went live.

Details:

This AI-driven enforcement marks a major step in combating fraud on a global scale, making ad platforms more secure. Google’s use of LLMs signals an increasingly important role for AI in detecting and preventing digital fraud. Despite the scale of these actions, Google emphasized its commitment to transparency, offering an appeal process for advertisers.

🚀 Boost your business with us—advertise where 10M+ AI leaders engage

🌟 Sign up for the first AI Hub in the world.

📲 Our Socials