🤖 OpenAI just launched the OpenAI Deployment Company.

A new enterprise focused unit to help companies deploy and scale AI into real world workflows.

- $4B initial investment
- 150 Forward Deployed Engineers via Tomoro acquisition

@aipost 🏴

🤖 OpenAI Introduces Daybreak: frontier AI for cyber defenders.

Daybreak brings together the most capable OpenAI models, Codex, and our security partners to accelerate cyber defense and continuously secure software.

A step toward a future where security teams can move at the speed defense demands.

Source.

@aipost 🏴

🔥 Unitree CEO Wang Xingxing just unveiled a real-life mecha.

Marketed as the world’s first mass-produced manned robot, this machine can transform into a quadrupedal civilian vehicle. The unit weighs roughly 500 kg (1,100 lb), including the pilot.

@aipost 🏴

🧠 Thinking Machines is replacing turn-based AI with always-present AI.

They just unveiled TML-Interaction-Small, a 276B-parameter MoE model built for continuous real-time interaction instead of stop-start conversations.

Most AI voice systems still work like walkie-talkies: you talk → they stop listening → they answer.

Thinking Machines slices audio, video, and text into 200ms micro-turns, letting the model:

• listen while speaking
• react to gestures and screen changes
• interrupt naturally when needed
• track silence and timing
• call tools in the background without leaving the conversation

The key difference is architectural: interactivity is trained into the model itself instead of patched together with separate voice detectors, speech systems, and timing rules.

Early results are impressive:

• 0.40s turn-taking latency
• 77.8 FD-Bench interaction score
• 43.4% on Audio MultiChallenge

This feels less like a chatbot waiting for prompts and more like AI that stays continuously present during interaction.

Source.

@aipost 🏴

🎓 Commencement speaker repeatedly gets booed by University of Central Florida students after she calls AI the 'next industrial revolution'

Students cheered when she said 'only a few years ago, AI was not a factor in our lives'.

The moment perfectly captured the growing divide around AI: tech leaders see a historic opportunity, while many students increasingly see uncertainty, disruption, and job anxiety.

AI may be advancing fast but public acceptance clearly isn’t moving at the same speed.

Full Speech.

@aipost 🏴

In November 2023, after Sam Altman had been fired, it was reported that OpenAI met secretly with Anthropic and asked Dario to merge Anthropic with OpenAI and then take over as CEO of the new company. And that Anthropic turned them down.

Source.

@aipost 🏴

This explains why Sam Altman might be so mad at Dario Amodei: during OpenAI’s messiest breakup episode, the board apparently tried to give Dario the keys, the house, and Sam’s job.

@aipost 🏴

🇺🇸 Microsoft CEO Satya Nadella just took the stand in Elon Musk’s OpenAI lawsuit.

Elon claims OpenAI abandoned its nonprofit mission and became a profit-driven company backed by Microsoft, despite the original vision he donated $38M toward.

Nadella pushed back, saying Microsoft knew nothing about any conditions tied to Elon’s donations and that OpenAI remained independent throughout.

Microsoft invested:

• $1B in 2019
• $2B in 2021
• $10B after ChatGPT launched

Back in 2020, Elon warned that OpenAI had been “essentially captured by Microsoft.”

Closing arguments are Thursday with Sam Altman expected to testify before then.

@aipost 🏴

Satya Nadella: “I don't want to be IBM and OpenAI to be Microsoft”

April 28, 2022.

@aipost 🏴

🇨🇳 DeepSeek just closed the biggest AI funding round in Chinese history but this is much bigger than a normal startup raise.

China is effectively turning DeepSeek into a national AI champion. After founder Liang Wenfeng was personally invited to a high-level meeting with Xi Jinping alongside leaders from BYD and Huawei, government support accelerated fast.

Now DeepSeek is reportedly raising $7B at a $51.5B valuation with backing tied to state-supported capital and Tencent.

But the real story is the rollout:

• 72 Chinese local governments are already using DeepSeek internally
• Shenzhen deployed it to 20,000 government workers
• Hospitals are using it for clinical support
• Universities run it on supercomputers
• Hotel chains and police departments are deploying it operationally

At the same time, China is pouring billions into AI infrastructure, subsidies, and compute access across the country.

The contrast with the US is striking: American AI companies compete for customers. China appears to be integrating one directly into the state itself.

@aipost 🏴

📈 The AI revolution is driving historic market momentum:

The S&P 500 excluding AI names is up +16% over the last 2 years and is trading below its February all-time high. Meanwhile, the S&P 500 as a whole is up +42% and has posted 15 all-time highs year-to-date. The S&P 500 has now surged for 6 consecutive weeks, its longest winning streak since October 2024.

Over this period, the index is up +16%, while the S&P 500 equal-weighted index is up +8%. This comes as 10 stocks have driven 69% of the S&P 500's gains during this time, with the remaining 490 representing just 31%. Alphabet, $GOOGL, NVIDIA, $NVDA, and Amazon, $AMZN, alone accounted for 15%, 10%, and 8%, respectively.

The AI race is accelerating.

@aipost 🏴

🎬 iQiyi says AI could produce most movies and TV shows within five years.

CEO Gong Yu unveiled Nadou Pro, an AI filmmaking toolkit for everything from scriptwriting to rendering calling it iQiyi’s biggest transformation in 16 years. The system uses models from Alibaba, ByteDance, and Google’s Veo 3.1.

iQiyi also plans to release 16 AI-made sci-fi and anime films, with its first commercial AI-powered movie expected this summer.

The push comes as the company faces an 81% profit drop, turning AI from an experiment into a survival strategy.

@aipost 🏴

Gemini Intelligence begins proactive AI rollout on Android

Gemini Intelligence is upgrading Android with new proactive AI capabilities that automate multi-step tasks across various apps, support browsing, and enable autofill functions.

This update marks a deeper integration into the Android operating system ahead of the anticipated revamp of Apple’s Siri, which is also expected to utilize Gemini technology.

The phased release is scheduled to start this summer with the newest Samsung Galaxy and Google Pixel devices. Additional Android platforms—including smartwatches, vehicles, smart glasses, and laptops—will follow later in the year. Further announcements are expected at next week’s industry events.

📰 @aipost

Japan’s Institute of Science Tokyo has launched a fully automated robotics laboratory equipped with 10 robots, including the Maholo LabDroid. These machines perform medical experiments such as handling reagents and cultivating cells.

Plans are underway to significantly expand the operation. The institute aims to deploy up to 2,000 research robots by 2040. The vision includes the integration of artificial intelligence to automate the entire research cycle, from generating hypotheses to verifying experimental results.

This initiative represents a substantial shift towards human-free laboratory environments, with the goal of increasing efficiency and accelerating the pace of scientific discovery.

📰 @aipost

❗️Today, nobody is against electricity, or electrification.

But the anti-data centre backlash we are seeing today has a precedent: the anti-electricity backlash of the early 1900s. They thought electricity would lead to horrible things.

Here's a propaganda cartoon from then.

@aipost 🏴

⚠️Top frontier AI labs now have very different model strengths.

• Claude Opus 4.7 is presented as the most consistently dominant model overall, ranking near the top across almost every major category.

• Gemini 3.1 Pro looks well-rounded with a creative writing edge.

• Muse Spark appears strong in overall performance and coding but weaker in expert, math, and long-query tasks.

• GPT-5.5 High is described as one of the most balanced models, especially strong in expert and math tasks.

• Grok 4.20 seems more specialized, standing out in creative writing and hard prompts.

No single lab owns every category, but Claude Opus 4.7 appears to be the most broadly consistent across the board.

@aipost 🏴

🧬 Isomorphic Labs raises $2.1B to build AI-designed medicines

Isomorphic Labs, the AI drug discovery company spun out of DeepMind, has raised a massive $2.1 billion round led by Thrive Capital with backing from Alphabet, GV, Temasek, and others.

The company is using AI to redesign how drugs are discovered, predicting promising molecules before they ever reach the lab, potentially cutting years off the traditional pharmaceutical process.

The funding will help Isomorphic scale its AI drug design engine, expand its treatment pipeline, and move more AI-created drugs toward clinical trials.

Its ambition is enormous: use frontier AI to eventually help “solve all disease.”

Source.

@aipost 🏴

🔥 Watch a team of humanoid robots run a full 8-hour shift at human performance levels.

Fully autonomous running Helix-02.

Watch here.

@aipost 🏴

I'm curious what happened here..

@aipost 🏴

🖱Google DeepMind is experimenting with an AI-powered mouse pointer that could change how people interact with AI entirely

The idea is simple: instead of writing long prompts, users could just point at something on their screen and give short commands through Gemini.

You might hover over a spreadsheet and say “make this a chart,” highlight text and say “summarize this,” or point at a recipe and say “double these ingredients.” The cursor itself becomes context for the AI.

It’s a subtle shift, but potentially a huge one. Today’s AI systems mostly rely on chat boxes and typed instructions. DeepMind’s experiments hint at a future where your mouse pointer acts like a real-time communication layer between you and the AI, helping the system instantly understand what you’re looking at, referring to, or trying to change.

Source.

@aipost 🏴

❗️OpenAI reportedly allowed more than 600 current and former employees to sell shares in a recent financing round, collectively cashing out about $6.6 billion.

Each person could sell up to $30 million worth of stock, with about 75 employees reportedly hitting that maximum. The sale happened before any IPO, showing how much private AI companies can now create life-changing wealth without ever going public.

@aipost 🏴