An AI-related provision in the “Big Beautiful Bill” could restrict state-level legislation of energy-hungry data centers—and is raising bipartisan objections across the US.
So-called “less-lethal” weapons like those that have been used against demonstrators in Los Angeles can cause severe, lasting harm like nerve or brain damage or blindness. They can also kill.
The FDA has issued its first ever approval on a safety consultation for lab-grown fish. That makes Wildtype only the fourth company to get approval from the regulator to sell cell-cultivated animal products, and its cultivated salmon is now available to order from one Portland restaurant.
Wildtype announced last week that the FDA had sent a letter declaring it had “no questions” about whether the cultivated salmon is “as safe as comparable foods,” the customary final step in the FDA’s approval process for lab-grown animal products. The FDA has sole responsibility for regulating most lab-grown seafood, whereas the task is shared with the United States Department of Agriculture (USDA) for cultivated meat.
The FDA’s pre-market safety consultation is voluntary, but is “helpful for marketability,” IP lawyer Dr. Emily Nytko-Lutz, who specializes in biotechnology patents, explained to The Verge. “There are other pathways involving self-affirmation of safety as well as a longer food additive review process, but the FDA’s authorisation with a ‘No Questions’ letter is a middle ground.”
Wildtype salmon is now on the menu at Haitian restaurant Kann in Portland, Oregon, and the company has opened a waitlist for the next five restaurants to stock the fish. It joins Upside Foods and Good Meat, two companies with permission to sell cultivated chicken in the US, while Mission Barns has been cleared by the FDA but is awaiting USDA approval for its cultivated pork fat. At a state level, the situation is more complicated, with eight states issuing bans on lab-grown meat as the technology becomes a conservative talking point.
Smart wearables company Ultrahuman has launched a new device that monitors changes in home environments that could impact your health. Ultrahuman says its $549 Home gadget tracks air quality, temperature, noise, light, and humidity, helping users optimize the climate within their homes to improve breathing and sleeping habits.
The Ultrahuman Home resembles a Mac Mini in terms of size and appearance. Its air quality features monitor levels of fine particulate matter, carbon monoxide, carbon dioxide, and chemical pollutants like acetone and formaldehyde. The device also tracks noise levels and various types of light exposure, including UVA, UVB, UVC, blue, red, green, and infrared, to help users “align home lighting with their body’s natural rhythm,” according to Ultrahuman.
Users who have an Ultrahuman Ring wearable can pair it with the Home device to unlock an “UltraSync” feature that suggests how environmental data may be impacting heart rate, sleep, and recovery patterns. For example, Ultrahuman says that UltraSync can suggest if the user was woken during the night by elevated noise or light levels.
We should note that the Ultrahuman Home won’t actually address the concerns it detects. The device is equipped with sensors and microphones for monitoring environmental changes via a mobile app, but it doesn’t include features like a built-in dehumidifier or air purification, and it doesn’t offer any way to integrate it into smart home ecosystems. There’s no recurring subscription to pay, and Ultrahuman says the “data and insights are with the user, always.”
Still, $549 is expensive for a device that doesn’t actually do anything — except maybe increase paranoia — unlike smart indoor air quality sensors available from Ikea, Amazon, SwitchBot and others.
With their quick-change camouflage and high level of intelligence, it’s not surprising that the public and scientific experts alike are fascinated by octopuses. Their abilities to recognize faces, solve puzzles, and learn behaviors from other octopuses make these animals a captivating study.
To perform these processes and others, like crawling or exploring, octopuses rely on their complex nervous system, one that has become a focus for neuroscientists. With about 500 million neurons—around the same number as dogs—octopuses’ nervous systems are the most complex of any invertebrate. But, unlike vertebrate organisms, the octopus’s nervous system is also decentralized, with around 350 million neurons, or 66 percent of it, located in its eight arms.
“This means each arm is capable of independently processing sensory input, initiating movement, and even executing complex behaviors—without direct instructions from the brain,” explains Galit Pelled, a professor of Mechanical Engineering, Radiology, and Neuroscience at Michigan State University who studies octopus neuroscience. “In essence, the arms have their own ‘mini-brains.’”
Galaxies are far more than the sum of their stars. Long before stars even formed, dark matter clumped up and drew regular matter together with its gravity, providing the invisible scaffolding upon which stars and galaxies eventually grew.
Today, nearly all galaxies are still embedded in giant “halos” of dark matter that extend far beyond their visible borders and hold them together, anchoring stars that move so quickly they would otherwise break out of their galaxy’s gravitational grip and spend their lives adrift in intergalactic space.
The way dark matter and stars interact influences how galaxies change over time. But until recently, scientists had mainly only examined one side of that relationship, exploring the way dark matter pulls on normal matter.
A robotic lander developed by a Japanese company named ispace plummeted to the Moon's surface Thursday, destroying a small rover and several experiments intended to demonstrate how future missions could mine and harvest lunar resources.
Ground teams at ispace's mission control center in Tokyo lost contact with the Resilience lunar lander moments before it was supposed to touch down in a region called Mare Frigoris, or the Sea of Cold, a basaltic plain in the Moon's northern hemisphere.
A few hours later, ispace officials confirmed what many observers suspected. The mission was lost. It's the second time ispace has failed to land on the Moon in as many tries.
In 2019, we told you about a new interactive digital "murder map" of London compiled by University of Cambridge criminologist Manuel Eisner. Drawing on data catalogued in the city coroners' rolls, the map showed the approximate location of 142 homicide cases in late medieval London. The Medieval Murder Maps project has since expanded to include maps of York and Oxford homicides, as well as podcast episodes focusing on individual cases.
It's easy to lose oneself down the rabbit hole of medieval murder for hours, filtering the killings by year, choice of weapon, and location. Think of it as a kind of 14th-century version of Clue: It was the noblewoman's hired assassins armed with daggers in the streets of Cheapside near St. Paul's Cathedral. And that's just the juiciest of the various cases described in a new paper published in the journal Criminal Law Forum.
The noblewoman was Ela Fitzpayne, wife of a knight named Sir Robert Fitzpayne, lord of Stogursey. The victim was a priest and her erstwhile lover, John Forde, who was stabbed to death in the streets of Cheapside on May 3, 1337. “We are looking at a murder commissioned by a leading figure of the English aristocracy," said University of Cambridge criminologist Manuel Eisner, who heads the Medieval Murder Maps project. "It is planned and cold-blooded, with a family member and close associates carrying it out, all of which suggests a revenge motive."
Everyone in quantum computing agrees that error correction will be the key to doing a broad range of useful calculations. But early every company in the field seems to have a different vision of how best to get there. Almost all of their plans share a key feature: some variation on logical qubits built by linking together multiple hardware qubits.
A key exception is Nord Quantique, which aims to dramatically cut the amount of hardware needed to support an error-corrected quantum computer. It does this by putting enough quantum states into a single piece of hardware, allowing each of those pieces to hold an error-corrected qubit. Last week, the company shared results showing that it could make hardware that used photons at two different frequencies to successfully identify every case where a logical qubit lost its state.
That still doesn't provide complete error correction, and they didn't use the logical qubit to perform operations. But it's an important validation of the company's approach.
A novel suggestion that complexity increases over time, not just in living organisms but in the nonliving world, promises to rewrite notions of time and evolution.
Beyond the jokes about its new shuttle service are serious questions about what it will mean for struggling transit systems, air quality, and congestion.
AI companies claim their tools couldn't exist without training on copyrighted material. It turns out, they could — it's just really hard. To prove it, AI researchers trained a new model that's less powerful but much more ethical. That's because the LLM's dataset uses only public domain and openly licensed material.
The paper (viaThe Washington Post) was a collaboration between 14 different institutions. The authors represent universities like MIT, Carnegie Mellon and the University of Toronto. Nonprofits like Vector Institute and the Allen Institute for AI also contributed.
The group built an 8 TB ethically-sourced dataset. Among the data was a set of 130,000 books in the Library of Congress. After inputting the material, they trained a seven-billion-parameter large language model (LLM) on that data. The result? It performed about as well as Meta's similarly sized Llama 2-7B from 2023. The team didn't publish benchmarks comparing its results to today's top models.
Performance comparable to a two-year-old model wasn't the only downside. The process of putting it all together was also a grind. Much of the data couldn't be read by machines, so humans had to sift through it. "We use automated tools, but all of our stuff was manually annotated at the end of the day and checked by people," co-author Stella Biderman told WaPo. "And that's just really hard." Figuring out the legal details also made the process hard. The team had to determine which license applied to each website they scanned.
So, what do you do with a less powerful LLM that's much harder to train? If nothing else, it can serve as a counterpoint.
In 2024, OpenAI told a British parliamentary committee that such a model essentially couldn't exist. The company claimed it would be "impossible to train today's leading AI models without using copyrighted materials." Last year, an Anthropic expert witness added, "LLMs would likely not exist if AI firms were required to license the works in their training datasets."
Of course, this study won't change the trajectory of AI companies. After all, more work to create less powerful tools doesn't jive with their interests. But at least it punctures one of the industry's common arguments. Don't be surprised if you hear about this study again in legal cases and regulation arguments.
This article originally appeared on Engadget at https://www.engadget.com/ai/it-turns-out-you-can-train-ai-models-without-copyrighted-material-174016619.html?src=rss
NEW BRAUNFELS, Texas—Abigail Lindsey worries the days of peace and quiet might be nearing an end at the rural, wooded property where she lives with her son. On the old ranch across the street, developers want to build an expansive complex of supercomputers for artificial intelligence, plus a large, private power plant to run it.
The plant would be big enough to power a major city, with 1,200 megawatts of planned generation capacity fueled by West Texas shale gas. It will only supply the new data center, and possibly other large data centers recently proposed, down the road.
“It just sucks,” Lindsey said, sitting on her deck in the shade of tall oak trees, outside the city of New Braunfels. “They’ve come in and will completely destroy our way of life: dark skies, quiet and peaceful.”
WASHINGTON, DC—The general outline of the Trump administration's proposed 2026 budget was released a few weeks back, and it included massive cuts for most agencies, including every one that funds scientific research. Late last week, those agencies began releasing details of what the cuts would mean for the actual projects and people they support. And the results are as bad as the initial budget had suggested: one-of-a-kind scientific experiment facilities and hardware retired, massive cuts in supported scientists, and entire areas of research halted.
And this comes in an environment where previously funded grants are being terminated, funding is being held up for ideological screening, and universities have been subjected to arbitrary funding freezes. Collectively, things are heading for damage to US science that will take decades to recover from. It's a radical break from the trajectory science had been on.
That's the environment that the US's National Academies of Science found itself in yesterday while hosting the State of the Science event in Washington, DC. It was an obvious opportunity for the nation's leading scientific organization to warn the nation of the consequences of the path that the current administration has been traveling. Instead, the event largely ignored the present to worry about a future that may never exist.
Over the years, scholars of the Dead Sea Scrolls have analyzed the ancient parchments with various methods: for example, X-rays, multispectral imaging, "virtual unfolding," and paleography, i.e., studying elements in their writing styles. The scrolls are believed to date back to between the third century BCE and the first century CE, but those dates rely largely on paleography, since only a handful of the scrolls have calendar dates written on them.
However, the traditional paleographic method is inherently subjective and based on a given scholar's experience. A team of scientists has combined radiocarbon dating from 24 scroll samples and machine-learning-based handwriting analysis to create their own AI program—dubbed Enoch. The objective was to achieve more accurate date estimates, according to a new paper published in the journal PLoS ONE. Among the findings: Many of the scrolls are older than previously thought.
As reported earlier, these ancient Hebrew texts—roughly 900 full and partial scrolls in all, stored in clay jars—were first discovered scattered in various caves near what was once the settlement of Qumran, just north of the Dead Sea, by Bedouin shepherds in 1946–1947. (Apparently, a shepherd threw a rock while searching for a lost member of his flock and accidentally shattered one of the clay jars, leading to the discovery.) Qumran was destroyed by the Romans, circa 73 CE, and historians believe the scrolls were hidden in the caves by a sect called the Essenes to protect them from being destroyed. The natural limestone and conditions within the caves helped preserve the scrolls for millennia.
Musk said in a post on X Thursday that SpaceX "will begin decommissioning its Dragon spacecraft immediately" in light of Trump's statement that floated canceling the billionaire's government contracts and subsidies.
The SpaceX CEO included a screenshot of Trump's earlier Truth Social post, which said terminating Musk's government contracts would be the "easiest way to save money in our Budget, Billions and Billions of Dollars."
Musk walked that decision back around five hours later.
"This is a shame this back and forth. You are both better than this. Cool off and take a step back for a couple days," X user Fab25june wrote on the platform.
"Good advice. Ok, we won't decommission Dragon," Musk said.
SpaceX's Dragon spaceships are used to transport NASA astronauts and supplies to and from the International Space Station.
In a statement to Business Insider, Bethany Stevens, NASA press secretary, said: "NASA will continue to execute upon the President's vision for the future of space. We will continue to work with our industry partners to ensure the President's objectives in space are met."
The White House did not respond to a request for comment.
The comments came as the feud between the former allies exploded on Thursday, with Trump and Musk publicly trading insults on their respective social media platforms, Truth Social and X.
Musk's government contracts are worth billions, with SpaceX working closely with NASA. SpaceX's Dragon spacecraft, designed to be reusable, can carry up to seven passengers to and from orbit and is the first private spacecraft to transport humans to and from the ISS, the company says.
Since 2020, NASA has relied on SpaceX's Dragon to transport astronauts to and from orbit. The agency, which retired its space shuttle program in 2011, depended on Russian Soyuz spacecraft for crewed missions prior to partnering with SpaceX.
In 2024, NASA announced SpaceX was awarded a $843 million contract to help decommission the ISS by the early 2030s. The plan involved using a larger, super-powered Dragon spaceship to push the ISS out of orbit, eventually landing in a remote part of the ocean. NASA planned to transition to using privately-owned space stations in the future.
Steve Bannon, who served as the White House chief strategist in Trump's first term, said in an interview Thursday that Trump should act immediately in response to Musk's announcement about decommissioning the Dragon spacecraft.
"President Trump tonight should sign an executive order calling for the Defense Production Act," Bannon said, referring to a federal law that grants the president authority to influence or control domestic industry in the name of national defense,"and seize SpaceX tonight before midnight."
Dublin AI startup Solidroad raises $6.5M from First Round Capital to transform customer service training with AI that coaches human agents and improves satisfaction scores.Read More
OpenAI reaches 3 million paying business users with 50% growth since February, launching new workplace AI tools including connectors and coding agents to compete with Microsoft.Read More
Mistral AI launches enterprise coding assistant with on-premise deployment to challenge GitHub Copilot, targeting corporate developers with data sovereignty and AI model customization.Read More
Login
Business Insider
VentureBeat
