News
Musk’s OpenAI will train artificial intelligence through video game ‘Universe’
Elon Musk’s OpenAI will introduce Universe, a virtual training ground aimed at teaching AI to play video games, use apps and even interact with websites. OpenAI, the artificial intelligence research company backed by the Tesla founder and billionaire entrepreneur, defines Universe in a blog post as “a software platform for measuring and training an AI’s general intelligence across the world’s supply of games, websites and other applications.”
Put simply, Universe will provide a gym that allows AI agents to go beyond their specialized knowledge of an individual environment to something approaching common sense. “Any task a human can complete with a computer.” Using a VNC (Virtual Network Computing) remote desktop, it allows the AI to control the game or app using a virtual keyboard and mouse, and to see its output by analyzing the pixels displayed on the screen. It’s essentially an interface to the company’s Gym toolkit for developing reinforcement algorithms, a type of machine learning system.
“Our goal is to develop a single AI agent that can flexibly apply its past experience on Universe environments to quickly master unfamiliar, difficult environments, which would be a major step towards general intelligence,” OpenAI says. As an example, it points to success of Google’s DeepMind AlphaGo initiative, which defeated the world champion human Go player earlier this year. While that success was impressive, when faced with a different challenge, the agent would have to go back to square one and learn the new environment through millions of trial and error steps.
OpenAI hopes to expand the Reward Learning (RL) lessons learned in one environment so that an AI agent can build upon past experience to succeed in unfamiliar environments.
We're releasing Universe, a platform for measuring and training AI agents: https://t.co/bx7OjMDaJK
— OpenAI (@OpenAI) December 5, 2016
OpenAI says in its blog post, “Systems with general problem solving ability — something akin to human common sense, allowing an agent to rapidly solve a new hard task — remain out of reach. One apparent challenge is that our agents don’t carry their experience along with them to new tasks. In a standard training regime, we initialize agents from scratch and let them twitch randomly through tens of millions of trials as they learn to repeat actions that happen to lead to rewarding outcomes. If we are to make progress towards generally intelligent agents, we must allow them to experience a wide repertoire of tasks so they can develop world knowledge and problem solving strategies that can be efficiently reused in a new task.”
-
Prior to Universe, the largest RL resource consisted of 55 Atari games — the Atari Learning Environment, says The Register. But Universe will begin with the largest library of games and resources ever assembled. “Out of the box, Universe comprises thousands of games (e.g. Flash games, slither.io, Starcraft), browser-based tasks (e.g. form filling), and applications (e.g. fold.it),” the OpenAI blog claims. Gaming companies that are cooperating with OpenAI include Flash, Microsoft – OpenAI announced a strategic partnership with the Redmond-based software giant – EA, Valve, Nvidia, Zachtronics, Wolfram, and others.
Universe is about more than gaming. It’s main focus is on training AI agents to complete common online tasks with speed and accuracy. “Today, our agents are mostly learning to interact with common user interface elements like buttons, lists and sliders, but in the future they could complete complex tasks, such as looking up things they don’t know on the internet, managing your email or calendar, completing Khan Academy lessons, or working on Amazon Mechanical Turk and CrowdFlower tasks.”
The OpenAI blog post introducing Universe gives a long and detailed accounting of how Universe was created and what it hopes to accomplish. At the end, it provides a number of ways that companies and individuals can contribute to the process. It’s fascinating reading for anyone interested in what the future of computing is likely to be.
There is also a darker side to artificial intelligence, which Elon Musk refers to as “summoning the Devil.” As The Register suggests, “While making software smarter may appeal to researchers, society as a whole appears to be increasingly unnerved by the prospect. Beyond the speculative fears about malevolent AI and more realistic concerns about the automation of military weaponry, companies and individuals already have trouble dealing with automated forms of interaction.”
One area of concern is that AI agents may one day be able to reactivate themselves after being shut down by human controllers. What was once the stuff of science fiction such as Minority Report and I, Robot could one day become all too real.
OpenAI Universe has been open-sourced on Github for those that may be interested in testing their own video game bot. We’ve included a video below showing OpenAI in action.
-
Tesla CEO Elon Musk has responded to reports that Volvo is officially discontinuing the LiDAR sensor on two of its cars.
Volvo announced that it would officially scrap LiDAR systems on its EX90 and ES90 vehicles in various markets. In Norway, owners will get a €1,800 compensation for features that will never arrive due to this decision. There will be no option to remove the unit from vehicles, either.
The issue stems from Volvo’s supplier, Luminar, and its bankruptcy filing. Luminar will no longer be able to supply LiDAR units to Volvo for the EX90 and ES90, effectively axing any use the unit has on vehicles. Volvo will phase out the data collection processes via the LiDAR system, and it will not be utilized whatsoever.
Musk saw the story on X and responded, stating:
“I did try to warn them. Humans drive using neural nets and optical sensors. Same is true for robot cars.”
I did try to warn them.
Humans drive using neural nets and optical sensors. Same is true for robot cars.
— Elon Musk (@elonmusk) July 29, 2026
-
Musk has been publicly vocal about his disdain for LiDAR systems, once calling them “a fool’s errand,” as he has consistently kept the outlook that they are not needed for effective self-driving.
The typical example used as evidence for this by Musk is humans themselves: made with only eyes and memories, humans are capable of navigating a car by using what they can see and what they’ve personally experienced on the road.
Elon Musk argues lidar and radar make self driving cars more dangerous
“Same is true for robot cars,” Musk says, as Teslas have eight exterior cameras that help see everything surrounding the vehicle, and a neural network that analyzes behavior and tendencies with every mile driven.
Tesla is a vision-only self-driving company that ditched sensors and radar several years ago in favor of cameras. Behind this effort, the company has established a reputation for having one of the most robust self-driving platforms in the world.
Musk’s big bet with Tesla on its self-driving program’s strategy has widely paid off. Other companies continue to utilize things like LiDAR, radar, and sensors for effective self-driving, but Tesla has shown that there is more than one way to give consumers a strong and accurate driver assistance suite.
-
The real question is: who will be the first company to take Musk’s advice and attempt a self-driving platform based on cameras only, or even license FSD for themselves?
Tesla has started to deploy the 2026 Summer Update to owners across its fleet, and among the biggest changes are improvements to Navigation, a new startup animation for the Model 3 and Model Y, Caraoke scoring, and new capabilities for Grok.
As the update has started making its way to some cars, we can now see a few of the features operating in real-time. We will show you what some of the new features look like in this article, along with some additional details on what changed.
Not all of the new features in the 2026 Summer Update have quite made an appearance, but some of them have, so we’ll show those here:
New Animation Screen for 3/Y
Tesla is rolling out a new startup animation for Tesla Model 3 and Model Y owners. This is present in Launch Edition and Performance Model 3 and Model Y, but other trim levels do not have anything like this.
The new animation screen on Model 3/Y with the 2026 Summer Update
Apparently color will also be customizable: pic.twitter.com/wSdwggtG3S
— TESLARATI (@Teslarati) July 27, 2026
-
Owners can adjust the color associated with the startup animation to suit their preferences. It is a surprise that more automakers do not focus on this animation for their vehicles; it can be a great first impression piece and make the car immediately feel more luxurious.
Tesla has included this on more premium trims, but it is nice to see it on the Model 3 and Model Y.
Grok Improvements
Grok can now adjust more things in the car outside of the Navigation system. Now, drivers can adjust anything from climate to driving settings by simply speaking to the AI assistant in the car:
@Tesla‘s AI assistant (Grok) with its new vehicle commands in action.
Set the climate, your destination and vehicle settings all in one go! pic.twitter.com/0bBTqaAk6G
— Not a Tesla App (@NotATeslaApp) July 28, 2026
You don’t even have to push a button, either. Instead, you can just say “Hey, Grok,” if you have it enabled. That feature rolled out with the 2026 Spring Update just a few months back.
-
This is a great feature, especially pertinent for the Robotaxi platform, as there will be no buttons inside the Cybercab when it eventually starts giving rides to the public. It also broadens Grok’s capabilities, which were relatively limited in terms of vehicle setting adjustments beforehand.
Tesla briefly offered one Robotaxi part in its Parts Catalog for your personal car, only to remove it just a short time after it was first noticed.
Tesla’s Robotaxi camera washer apparatus was briefly available for purchase on the company’s Online Parts Catalog. The camera washer was first noticed on Model Y Robotaxi vehicles about six months ago in Austin.
🚨 Tesla looks to have installed Camera Washers on the side repeater cameras on Robotaxis in Austin
pic.twitter.com/xemRtDtlRR— TESLARATI (@Teslarati) January 23, 2026
First noticed by Not a Tesla App, the Camera Washer entries appeared for the new “Juniper” Model Y under a category called “Halo,” which has also now disappeared. Interestingly, Halo probably is related to Tesla’s internal “Project Halo,” which was a project that aimed to retrofit customer-owned Model Ys into functional Robotaxis.
This hardware addition would likely be required for the vehicle to operate as a Robotaxi, as the Camera Washer seems to be a non-negotiable part of the vision-based system Tesla utilizes for self-driving efforts.
However, this part has since been removed and is no longer visible on the EPC.
-
Now the true question lingers: Why would Tesla add this Camera Washer to the Model Y parts catalog? Is it planning to make it available for owners to utilize on their own cars for personal use, or will it become a prerequisite for Robotaxi operation in customer-owned cars?
While discussing the upgrade options for Hardware 3 vehicles during the Q1 Earnings Call, Tesla CEO Elon Musk had said that the company could establish small, satellite shops that would upgrade cameras and self-driving computers. Perhaps this same strategy could be utilized for vehicles that want to be included in Robotaxi but do not have the correct hardware.
AI4 is currently represented as capable of unsupervised self-driving, and the same was said about HW3 at one point, only for Tesla to admit last quarter that it would, unfortunately, not be possible. Perhaps AI4 vehicles might need this camera washer as a prerequisite, just as HW3 cars will need that camera and computer upgrade.
This could be the first hint of that’s where we are headed.
Elon Musk responds to Volvo’s latest LiDAR decision
Tesla unlocks in-car web conferencing to extend Google Meet, Teams and beyond
