Those with an interest in the concept of AI alignment (i.e., getting AIs to stick to human-authored ethical rules) may remember when Anthropic claimed its Opus 4 model resorted to blackmail to stay online in a theoretical testing scenario last year. Now, Anthropic says it thinks this "misalignment" was primarily the result of training on "internet text that portrays AI as evil and interested in self-preservation."

In a recent technical post on Anthropic's Alignment Science blog (and an accompanying social media thread and public-facing blog post ), Anthropic researchers lay out their attempts to correct for the kind of "unsafe" AI behavior that "the model most likely learned... through science fiction stories, many of which depict an AI that is not as aligned as we would like Claude to be." In the end, the model maker says the best remedy for overriding those "evil AI" stories might be additional training with synthetic stories showing an AI acting ethically.

"The beginning of a dramatic story..."

After a model's initial training on a large corpus of mostly Internet-derived data, Anthropic follows a post-training process intended to nudge the final model toward being "helpful, honest, and harmless" (HHH). In the past, Anthropic said this post-training has leaned on chat-based reinforcement learning with human feedback (RLHF), which it said was "sufficient" for models used mostly for chatting with users.

Read full article

Comments

For decades, astronomers looking through telescopes like Hubble have been trying to catch a glimpse of the ancient epoch when the Universe's first generation of stars ignited. But the small galaxies that were the building blocks of the cosmos we know today were too faint to spot, even by the most powerful instruments. Now it seems astronomers finally have two things on their side: the Webb Space Telescope and a bit of luck.

In a recent paper in Nature, a team of scientists led by Kimihiko Nakajima, an astronomer at the Kanazawa University, Japan, used the James Webb Space Telescope to observe an ultra-faint galaxy called LAP1-B as it existed roughly 800 million years after the Big Bang. It’s the most chemically primitive galaxy we’ve ever seen.

The magnifying glass

The LAP1-B is 13 billion light-years away from Earth. To observe an object that faint and distant, even the huge, gold-coated beryllium mirrors of JWST were not enough on their own. We spotted it due to a massive cluster of galaxies called the MACS J046, which warps the spacetime between us and the LAP1-B.

Read full article

Comments

Blue Origin is weighing its first external fundraising as part of a push by Jeff Bezos’ rocket venture to hit ambitious launch targets and tap investor appetite boosted by SpaceX’s upcoming initial public offering.

Chief Executive Dave Limp told employees at a recent all-hands meeting that the company would require outside investment if it were to significantly increase its launch cadence, according to details of the meeting from two people who attended.

He said it would “take a lot of capital” to achieve the number of rocket launches Blue Origin has targeted—more money than would be available with “just one investor,” the people added.

Read full article

Comments

We loved the first season of Daredevil: Born Again , Marvel's hotly anticipated revival of the popular series in the Netflix Defenders universe, and its sophomore outing did not disappoint. The show just wrapped its critically acclaimed second season, with a third already well underway—all part of MCU's Phase Six master plan.

(Some spoilers below, but we'll give you a heads up before any major S2 reveals.)

From its inception, Daredevil: Born Again was built around the conflict between Matt Murdock/Daredevil (Charlie Cox) and Wilson Fisk/Kingpin (Vincent D'Onofrio), with Fisk attempting to leave his criminal past behind as the newly elected mayor of New York, and Murdock determined to abandon his vigilante activities as Daredevil to focus full time on his law practice.

Read full article

Comments

Rivian has quickly built a reputation as one of the auto industry's leaders when it comes to vehicle software. Its clean-sheet approach to an electric vehicle's electronic architecture earned it a $5 billion investment from Volkswagen Group, and its in-house infotainment system is beloved by owners despite no plans inside the company to support phone mirroring through Apple CarPlay or Android Auto.

In the absence of phone mirroring—and the way it lets you easily use Siri or Google Assistant hands-free while driving—Rivian has now added a new AI digital helper in its latest software update, compatible with both older Gen1 Rivians (model-year 2024 and older) as well as the more recent Gen2 models .

The Rivian Assistant rolled out in its latest software update , 2026.15, to all owners with a subscription or trial for Connect+, Rivian's connectivity services. You activate it like most digital assistants, either with a button on the steering wheel, an icon on the infotainment display, or with a trigger phrase—in this case "Hey Rivian" or "OK, Rivian."

Read full article

Comments

NASA has enabled scientists to study the impact of microgravity on drug development for decades, beginning with the Space Shuttle. This work accelerated in the 2010s, with the completion of the International Space Station and full-time crew members devoted to scientific research.

There have been some notable successes during this timeframe, such as the ability to grow a more uniform crystalline form of the cancer drug Keytruda in 2019 . This opened up the possibility of administering the drug via injection rather than requiring a patient to spend hours in a clinic setting to receive the drug intravenously.

NASA subsidized much of this work, typically paying the considerable costs to transport research to the ISS and for astronaut time to conduct research there. There were, however, trade-offs, such as long lead times to get research into space. Nevertheless, it has become clear that there could be some commercial applications for making drugs in space.

Read full article

Comments

Data centers may be coming to your neighborhood as side installations associated with new homes—and in exchange would offer subsidized electricity and Internet access along with backup batteries to homeowners. The company behind the plan has already begun pilot testing in preparation for a 100-home trial run this year.

The “distributed data center solution” announced by the San Francisco startup SPAN would deploy thousands of XFRA nodes that contain liquid-cooled Nvidia RTX Pro 6000 Blackwell Server Edition GPUs operating with minimal noise, according to a press release. By harnessing excess power capacity among US households, SPAN aims to quickly expand the available compute for AI workloads without the costs and delays associated with trying to build warehouse-sized data centers.

“Data centers are loud, ugly, and often drive up local electricity bills,” said Chris Lander, vice president of XFRA at SPAN, in correspondence with Ars. “[This] is quiet, discreet, and makes energy more affordable for the host and community.”

Read full article

Comments

Marty Makary on Tuesday resigned from his role as Commissioner of the Food and Drug Administration, days after news broke on Friday that the White House had signed off on plans to fire him .

Trump confirmed Makary's resignation on social media, posting an image that appears to show that Makary resigned from his role over a text message. The text message begins "Dr. President Trump[sic], Please accept my resignation, effective today."

Trump wrote in another social media post that Makary had "done a great job at the FDA," and that he was "a hard worker, who was respected by all, and will go on to have an outstanding career in Medicine."

Read full article

Comments

In the US, fired and laid-off workers often have their digital credentials deactivated before they learn about the loss of their jobs; indeed, the inability to log in to a corporate system may be the first an employee knows of the situation.

Although not a generous or humane approach to staff reduction, it does follow from the simple fact that a fired employee with access to company systems is a security risk.

Just ask the Akhter twin brothers, accused of wiping out 96 databases hosting US government information in the minutes after both were fired last year from their shared employer.

Read full article

Comments