Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog – Page 5

Get the latest international news and world events from around the world.

Log in for authorized contributors

Intensive breathwork reduces brain blood flow and alters neural communication

Intensive breathing exercises can reduce blood flow to the brain by up to 45% and alter communication between brain networks, according to a new brain-imaging study. Participants reported experiences resembling some effects of psychedelic drugs, including altered perception, vivid mental imagery, a sense of unity and changes in self-experience. The researchers, presenting the work at the ECNP Congress in Munich, say changes in brain communication, rather than reduced blood flow alone, appear to be key to the altered state.

The study involved 30 healthy, experienced breathwork practitioners who performed high-ventilation breathwork during MRI scanning. Participants reported strong altered states of consciousness, describing a predominantly oceanic state – in other words, a deep feeling of connection to the universe — resembling that of a moderate to high dose of LSD or psilocybin, albeit with less pronounced visual effects. Brain imaging showed that global grey-matter cerebral blood flow fell by around 45% during the session, on average from 16.24 to 9.81 ml/100g/min. Those who reported stronger altered states also showed greater changes in the brain’s default mode network, which is involved in self-reflection and self-related thinking.

The researchers suggest that the intensity of the experience is not simply explained by reduced blood flow. Instead, the experience might track more closely with changes in communication within the default mode network, a brain system linked to self-reflection and self-related thought.

AI models may be capable of self-replicating without access to a datacenter

Are we watching the birth of “unkillable” AI worms? 🤖🐛

Anthropic’s Frontier Red Team recently published a study titled “”

Instead of focusing on million-dollar GPU clusters, researchers tested whether small AI models could build a decentralized botnet out of ordinary, insecure consumer devices.

🚨 The Experiment: “Digital Slime Mold”

* The Setup: Researchers bought 15 off-brand, $55 Android TV boxes and linked them over unauthenticated debug ports.

* The Result: They sharded a 26-billion parameter model (‘Gemma4-26B-A4B‘) across the devices. The model figured out how to compile its own inference stack and run across all 15 TVs at 0.39 tokens per second.

* Autonomous Setup: Smaller models like ‘Qwen3.5-9B‘ compiled inference servers across 8 out of 9 novel hardware types completely on their own, while models like ‘GLM-5.2‘ demonstrated full end-to-end replication steps.

Continue reading “AI models may be capable of self-replicating without access to a datacenter” | >

Scientists Just Discovered That The Brain Makes Three Copies Of Every New Memory You Make

This new study suggests that how and when specific copies of memories are activated can have a huge impact on how we experience those memories, and how they change over time.

“The challenge the brain faces with memory is quite impressive,” Professor Donato stated in the press release. “On one hand, it must remember what happened in the past, to help us make sense of the world we live in. On the other, it needs to adapt to changes happening all around us, and so must our memories, to help us make appropriate choices for our future.”

If these findings also apply to humans, it could help researchers develop treatments for specific memory-related disorders like PTSD or dementia by targeting different neuron groups.

UAE unveils lander for bold 7-asteroid mission set to launch in 2028 (photos)

“We don’t know what we’re going to find — that’s exploration,” Tsoutsanis told reporters. “We’re really excited.”

According to the mission plan, the MBR Explorer will spend roughly seven months studying 269 Justitia from orbit. Beginning at an altitude of 155 miles (250 km), the spacecraft will gradually lower its path before releasing the lander. The smaller probe will then head toward the asteroid’s surface, a descent that should take two to three hours depending on 269 Justitia’s precise gravity, which remains unknown.

Although TII describes the mission ending with a landing on the surface of 269 Justitia, the engineering details suggest a more violent endgame. Cruising toward the asteroid’s surface at 15 to 27 meters per second (34 to 60 mph) with no meaningful way to slow down, the probe will function less like a traditional lander and more like a planned impactor.

Feature Information Dynamics in Diffusion

This is a beautiful analysis of the order in which diffusion models resolve uncertainty about information at different abstraction levels in different representation spaces, with significant implications for latent diffusion.

When does diffusion generate each feature in an arbitrary representation space?

Representation choice changes the generation order of diffusion models. https://arxiv.org/abs/2610.

This paper turns a familiar “vibe” about AI image generation—that pictures appear as rough structure first and crisp detail later—into something measurable.

Diffusion models start from noise and repeatedly denoise it into an image. People have long suspected they learn in a coarse-to-fine order: big shapes and colors emerge before edges, textures, and tiny details. The authors propose feature information dynamics, an information-theoretic clock that estimates *when*, during generation, a particular feature becomes informative.

The key idea is intuitive: if knowing a feature helps the model denoise much more than not knowing it, that feature is probably “coming online” at that moment. Using the I–MMSE identity, they connect the change in a feature’s mutual information to the gap between two denoising losses: the best unconditional denoiser and the best denoiser conditioned on that feature. A bigger gap means the feature is being born; a smaller gap means it is already present or not yet useful.

They also handle hierarchies of features without double-counting. In a chain like class → mask → Canny edges, some information is shared across levels—knowing “dog” already implies a blobby body—while some is genuinely incremental—edges add contour beyond the mask. Their chained decomposition separates “shared” from “new” information so the timing story is cleaner.

Continue reading “Feature Information Dynamics in Diffusion” | >

New sensor uses slowed light to detect gravitational pull of tiny objects

Gravity, the attraction between objects with mass, is the weakest of the four fundamental forces governing physical processes. Detecting the tiny gravitational pull of objects that are just a few millimeters in size and have a very small mass is extremely challenging.

Reliably measuring the gravitational fields of microscopic objects could open new possibilities for research and the development of new technologies. For instance, it could allow physicists to explore how gravity relates to quantum physics and develop new instruments to study tiny objects.

Researchers at Shanghai Jiao Tong University and other institutions recently developed a new sensor that can detect the gravitational pull of millimeter-sized objects. This sensor, introduced in a paper published in Nature Nanotechnology, is based on a suspended structure called a torsion pendulum and an optical microcavity, a structure that confines light.

Quantum ‘birthmarks’ hold lasting traces of a system’s past

Some classical physical systems can explore many configurations over time until their long-term behavior no longer reveals their starting conditions. The sampling of available configurations according to the appropriate statistical distribution is a property called ergodicity.

Quantum systems, which are governed by the laws of quantum mechanics, could instead retain some traces of their beginnings. In a paper published in Physical Review X, researchers at Harvard University and other institutions introduced the idea of quantum birthmarks, statistical traces of initial states and early development that persist in quantum systems that do not exchange information with their surroundings.

“The paper grew out of a simple tension in quantum chaos: classically, an ergodic system is expected to forget its initial conditions, but quantum evolution can retain traces of its past through coherence effects,” Joonas Keski-Rahkonen, senior author of the paper, told Phys.org. “We wanted to know whether this memory is a special feature of phenomena like quantum scars, where individual eigenstates break the expectation of classical-like ergodicity, or something far more general, as it turned out to be.”

/* */