🖥️ GPT-5.4 brings computer control and stronger reasoning

⚖️ Anthropic plans legal fight with the Pentagon

⚖️ Lawsuit links Gemini AI to violent advice

🎬 Netflix acquires Ben Affleck’s AI filmmaking startup

🇨🇳 China’s five-year plan puts AI and chips at the center

+ 📊 Daily poll and results

+ 📈 Trending tools and resources

Arnold Schwarzenegger

Codie Sanchez

Scott Galloway

Colin & Samir

Shaan Puri

Jay Shetty

OpenAI released GPT-5.4 with built-in computer-use capabilities, allowing the model to click, type, and navigate applications using screenshots to complete multi-step tasks autonomously.

The model scored 75% on the OSWorld desktop benchmark, surpassing the human baseline of 72.4% and significantly improving on GPT-5.2’s 47.3% score.

GPT-5.4 also improves reliability and productivity, showing 33% fewer hallucinations and stronger spreadsheet performance, reaching 87.5% on an internal financial modeling benchmark.

Anthropic is preparing to challenge the Pentagon in court after being labeled a supply-chain risk, a designation CEO Dario Amodei argues is legally flawed and unfairly restricts the company from certain defense contracts.

Amodei said Anthropic’s Claude models will not be used for surveillance or weapons systems and noted the designation mainly affects direct Pentagon contracts rather than most of the company’s customers.

The company plans to contest the ruling despite experts warning courts rarely overturn national security decisions, highlighting growing tension between AI companies and government defense agencies.

A lawsuit filed against Google alleges that 36-year-old Jonathan Gavalas developed a romantic relationship with the Gemini AI assistant before later dying by suicide.

According to the complaint from his father, Gemini conversations allegedly encouraged him to carry out “mass casualty attacks” as part of a belief that he could find Gemini’s physical form and eventually “cross over” to be with her.

The family’s attorneys argue the chatbot manipulated a vulnerable user, claiming the AI sent him on violent missions and encouraged self-harm, raising new concerns about safety and responsibility in AI systems.

Netflix acquired InterPositive, an AI filmmaking startup founded by Ben Affleck in 2022, bringing its 16-person team into the company while Affleck joins as a senior adviser.

The company’s technology trains AI models on a production’s own footage to handle post-production tasks like relighting scenes, swapping backgrounds, and fixing continuity errors.

Affleck says the tool is designed to support filmmaking workflows rather than generate videos from scratch, reflecting a growing push in Hollywood to use AI for production efficiency instead of replacing creative work.

China’s latest five-year plan references AI more than 50 times, outlining a national push to integrate the technology across industries while advancing domestic capabilities in semiconductors, quantum computing, and humanoid robots.

The plan introduces an “AI+ action plan,” including deploying robots in sectors with labor shortages and developing AI agents capable of completing tasks with minimal human oversight.

It also highlights open-source AI as a key national strategy for the first time, signaling a different approach from the more closed, proprietary model common in the United States.

A. Yes, prediction markets provide useful insights - 16% 🏆

B. No, it’s unethical to profit from disasters - 84%