SUMM

Thinking Machines raised $2 billion led by A16Z, with backing from Nvidia, AMD, Cisco, Service Now, Excel, Jane Street, and others.
The company aims to build multimodal AI for natural human-like interaction and collaboration, targeting enterprise applications.
They promise a forthcoming project with open-source elements, useful for researchers and startups, and plan to share scientific findings about frontier AI systems.
The creator expresses skepticism about AI companies’ open-source claims but hopes Thinking Machines will deliver.

Whimo achieved 100 million autonomous rides, operating in just five US cities.
The host shares a positive experience with Whimo, noting human-like and even superior decision making.
The rapid growth suggests the company could easily reach higher ride milestones.

Sam Altman echoed Nvidia CEO Jensen Huang’s optimism about AI and the job market.
He predicts that as the cost of intelligence drops, more use cases will emerge, boosting productivity and expanding opportunities.
Human creativity, taste, and the drive for novel experiences will continue to matter even as AI advances.
People will still want to create and be useful, and new types of jobs will emerge, potentially as meaningful as today’s creative digital roles.
The speaker sides with an optimistic view of humans and AI working together for a better future.

Uber is reentering autonomous driving after scrapping its previous program following a fatal accident.
The company has partnered with Lucid Motors to develop robo-taxis, exclusively for Uber, with a $300 million investment.
The market is dominated by data-rich players like Tesla and Whimo, making competition difficult, but new efforts are welcomed.

Meta has aggressively hired key AI talent from Apple, including two researchers from Apple’s super intelligence labs, after previously poaching their boss.
Meta is offering substantial compensation packages for AI researchers.
Apple is seen as lagging behind in AI deployment and features; the speaker expresses disappointment with Siri and Apple Intelligence.
Microsoft has recruited over 20 AI employees from Google DeepMind, including the former head of engineering for the Gemini chatbot.

OpenAI designated its new ChatGPT Agent as high risk under its preparedness framework due to potential biosecurity misuse.
The agent autonomously browses the web, takes actions, and can complete extended tasks, raising unique safety concerns.
Strong safeguards have been activated, including threat modeling, dual-use refusals, classifiers, monitoring, red teaming by international experts, and a biobug bounty program.
Despite precautions, the speaker notes no model is perfect and vulnerabilities may still be exposed.

DeckArt introduced Mirage LSD, the first live stream diffusion model capable of real-time video transformation at 24 fps and under 40ms latency, using novel efficiency gains.
South Korea’s Upstage AI launched Solar Pro 2, a 31B parameter hybrid reasoning model, competitively priced at $0.50 per million tokens, currently outperforming many peers on certain benchmarks.
The Ark Prize launched ARC AGI 3, a new interactive benchmark focused on puzzle-solving with minimal instruction, designed to be easy for humans but difficult for AIs; current top model scores are around 16–17%.

Stargate, a massive compute infrastructure partnership between SoftBank, OpenAI, Oracle, and the US government, has reportedly scaled back initial plans and has not completed a data center deal yet.
Despite setbacks, Oracle and OpenAI recently announced plans for an additional 4.5 GW of data center capacity in the US, projecting the creation of 100,000 jobs.
The combined effort aims for 10 GW of new capacity, positioning Stargate as one of the largest AI infrastructure projects in US history.

OpenAI is testing a new feature allowing users to select from different ChatGPT personalities, such as skeptic, robot, listener, sage, and default.
The host notes the feature was not yet live in their account but is being gradually rolled out.

A leaked memo reveals Anthropic’s intention to seek investment from UAE and Qatar, spurred by the vast capital available in the Middle East (over $100 billion).
The memo discusses the ethical dilemma of working with entities considered unsavory to stay competitive with funding.
Anthropic maintains opposition to exporting powerful AI infrastructure to authoritarian regimes, citing national security and the risks of increased technological leverage.

The video is sponsored by Augment Code, an AI assistant for professional software engineers, designed for large codebases and providing advanced contextual suggestions and remote automation.
The host encourages viewers to try Augment Code and promotes the video’s channel.

AI News: Sam Altman's Predictions, Talent Wars Continue, Project Stargate, Thinking Machines