Agregátor RSS

Ink/Stitch byl vydán ve verzi 3.1.0. Jedná se o open source software, rozšíření Inkscapu, pro návrh výšivek pro vyšívací stroje.

Open-source pixelartový editor Pixelorama napsaný v enginu Godot dospěl do verze 1.0. Přidává řadu nových nástrojů a používá nový formát souborů, který není zpětně kompatibilní s dřívějšími verzemi. Aplikace je multiplatformní, běží také v prohlížeči a nově se dá vedle itch.io zakoupit i na Steamu.

OpenTofu, tj. svobodný a otevřený fork Terraformu vzniknuvší jako reakce na přelicencování Terraformu z MPL na BSL (Business Source License) společností HashiCorp, bylo vydáno ve verzi 1.8.0. Přehled novinek v aktualizované dokumentaci.

NÁZORY A KOMENTÁŘE. Povinná registrace ubytovacích zařízení v novém systému eTurista se bude týkat i lidí, kteří si přivydělávají. Třeba jen v létě přenecháváním bytu či pokoje v rodinném domku, chaty k ubytování rekreantům. Ani je však registrace a další povinnosti neminou.

Mezníkem ve vývoji platformy PC byl rok 1987, protože právě v tomto roce začala být prodávána grafická karta VGA. Umožnila vývoj sofistikovanějších her a později byla kombinace VGA+Intel 80386 použita ve hře Doom.

Magie v tom sice není, ale jména lidí mají přesto ohromnou moc. Podle izraelských vědců dokážeme poměrně dobře přiřadit tvář neznámého dospělého člověka ke jménu. U dětí to ale nefunguje. Vypadá to, že sociální očekávání, která se pojí se jménem, mohutně tvarují vzhled dotyčného člověka. Jak jste na tom se jménem vy?

Fosilie Argonautů v místní Chóře. Majáček nad antickým přístavem. Posvátná cesta horským svahem od antického města k Apollónovu chrámu, místy s původní dlažbou. Chrám „na konci světa“, dávno přestavěný na klášter, naštěstí hodně zajímavě a s ponecháním opěrných zdí celého okrsku. Dál je už jen Dračí sluj a šílený útes Kalamos. To vše na ostrově Anafi.

Informace, které v souvislosti s kauzou kolem odcházejících procesorů Intelu vyplouvají na povrch, ukazují na souhru řady problémů, které vedly k současnému výsledku…

'Ignore previous instructions' thwarts Prompt-Guard model if you just add some good ol' ASCII code 32

Meta's machine-learning model for detecting prompt injection attacks – special prompts to make neural networks behave inappropriately – is itself vulnerable to, you guessed it, prompt injection attacks.…

The brain is like a medieval castle perched on a cliff, protected on all sides by high walls, making it nearly impenetrable.

Its shield is the blood-brain barrier, a layer of tightly connected cells that only allows an extremely selective group of molecules to pass. The barrier keeps delicate brain cells safely away from harmful substances, but it also blocks therapeutic proteins—like, for example, those that grab onto and neutralize toxic clumps in Alzheimer’s disease.

One way to smuggle proteins across? A cat parasite.

A new study in Nature Microbiology tapped into the strange world of mind-bending parasites, specifically, Toxoplasma gondii. Perhaps best known for its ability to rid infected mice of their fear of cats, the parasite naturally travels from the gut to the brain—including ours—and releases proteins that tweak behavior.

The international team hijacked T. gondii’s natural, brain-targeting impulses to engineer two delivery systems, one for a single-shot therapeutic boost and another that lasts longer.

The unconventional shuttle worked on brain cells in petri dishes and brain organoids. Often called “mini-brains,” these pea-sized blobs roughly capture the cell types and structure of a growing fetal human brain. However, they don’t usually produce a blood-brain barrier.

To show the shuttle could gain access to the brain, the team engineered a T. gondii shuttle with a therapeutic protein for Rett syndrome, a genetic disorder that leads to autism-like symptoms.

After one shot into the belly, the shuttle released the therapeutic proteins widely into the brains of lab mice within a few weeks. The proteins mostly accumulated in parts of the brain critical for perception, reasoning, and memory.

“For medicine, efficient and safe delivery of proteins could unlock a broad category of protein-based therapies,” wrote the authors.

U-Haul to the Brain

Getting protein-based drugs into the brain is a pain. Unlike gene therapy concoctions, proteins are extremely sensitive to heat and acid. They can’t be swallowed as a pill—the gut’s acid destroys them. Even injections straight into the blood stream are problematic. Immune cells, for example, may wipe out the proteins before they have a chance to reach the brain.

Thankfully, nature is a source of inspiration. All brain-targeting carriers need to bypass two “checkpoints”: The first is the blood-brain barrier, the second, the neuron’s membrane.

A popular approach uses a bio-engineered virus carrying the genetic instructions to make a protein once inside the neurons. Often employed in gene therapy, scientists make the virus relatively safe by stripping away its infectious tendencies. But like a small U-Haul van, it only has room for the genetic instructions of smaller proteins.

Another surprising carrier traces its roots to HIV. Scientists studying the virus found a small protein chunk that allows it to penetrate the blood-brain barrier and get past neuron membranes. By engineering these chunks—which aren’t infectious—into shuttles, scientists can then tag protein cargos onto them. One example (by yours truly) could tunnel into the brain after an injection into the bloodstream and protect rats’ brains from damage after a stroke.

These shuttles too are limited by size: They can only drag along very small protein snippets. Antibodies and other larger proteins are beyond reach.

T. gondii, in contrast, has a much larger capacity.

A Synthetic Fleet

A cat parasite hardly sounds like medicine. But it’s a worthy candidate.

Normally, T. gondii produces egg-like “offspring” in the guts of cats, which are then strewn into the wild as they poop. The parasite waits for potential hosts—say, a mouse sniffing for crumbs or a human changing the litter box—and infects the unsuspecting host, ultimately spreading into the brain. Once inside, T. gondii lingers in neurons, rather than other brain cells.

It sounds terrifying, but for people with a healthy immune system, the parasite usually doesn’t cause harm. “In fact, it is estimated that a third of the world population is chronically infected with the parasite,” Dr. Oded Rechavi’s lab, who led the study, wrote in a blog post.

To transform T. gondii into a delivery tool, the team focused on two secretion systems in the parasite that let the parasite pump proteins into target cells. These are “remarkable innate abilities,” wrote the team.

They first built a protein link between the two systems and their potential cargo, for example, proteins implicated in Parkinson’s disease, gene-editing proteins, and MECP2—which is linked to Rett syndrome. The team then tethered the proteins to one of the two systems and delivered them into a variety of cells in petri dishes.

Within a day, the proteins were thriving inside their hosts.

In neurons without MECP2, a dose of T. gondii carrying a synthetic version of the protein boosted its levels to roughly 58 percent of normal cells, which is similar to previous gene therapy studies of Rett syndrome. The added MECP2 worked like its natural counterpart, turning genes on or off inside neurons as expected.

T. gondii also reliably released its payload into mature brain organoids. The protein altered genetic transcription throughout the mini-brains, changing gene expression as predicted.

The two T. gondii systems had individual strengths. One is a “kiss-and-spit”: Like a fighter jet, T. gondii swoops in on a neuron, releases its protein payload, and leaves. The other takes a longer approach, requiring T. gondii to infiltrate and establish itself inside the cell, like a sleeper agent. Once in, however, the system can deliver its cargo for a longer time and at a higher level.

Cat and Mouse Game

As a final test, the team injected the engineered T. gondii, with an MECP2 payload, into the bellies of mice—like an insulin shot for people with diabetes.

Eighteen days later, the mice’s brains showed signs of cysts—which are harmless for people without immune problems—indicating the parasite was establishing itself inside the brain. Other tissues, including the liver, lung, and spleen, had very little T. gondii roaming around for up to three months after injection. Only the brain had a boost in MECP2.

“Many proteins require controlled targeting” to a specific part of the body, or otherwise they’re “ineffective or even deleterious if delivered elsewhere,” explained the team.

Surveying multiple regions of the brain, T. gondii seemed to prefer settling inside the cortex—the outermost region of the brain involved in perception, reasoning, and making decisions. Its second choice was the “memory center,” the hippocampus. That’s good news: Both regions are favorite targets for tackling neurological disorders. And the treatment didn’t alert the body’s immune system, with the therapeutic proteins easily getting along with the brain’s usual protein brigade.

“T. gondii can be used…[for]…many of the challenges associated with protein delivery,” for both scientific research and therapeutics, wrote the team.

There’s still a long road to go. Although T. gondii is safe for healthy people, it has been linked to side effects in the brain for the immunocompromised. The next step is to strip away its toxicity in a way similar to the viral carriers now used for gene therapy. If it works, T. gondii is set for a genetic makeover as a safe shuttle to the brain—despite its cat parasite origin story.

Image Credit: T. gondii cyst in mouse brain tissue. Jitinder P. Dubey / Wikimedia Commons

Do we really want to bother SCOTUS with this, friends? Surely they're way too busy to take a look

US border agents must obtain a warrant, in New York at least, to search anyone's phone and other electronic device when traveling in or out of the country, another federal judge has ruled.…

Dáte na doporučení výrobců tiskáren a kupujete pouze ty jejich originální náplně? Nebo raději ušetříte a vyberete alternativní inkoustové kazety, lahvičky či tonery? Vedle hlasu v anketě můžete také napsat, jakou tiskárnu používáte, jak často tisknete a jestli jste někdy pozorovali rozdíl v ...

Byla vydána nová verze 2.46.0 distribuovaného systému správy verzí Git. Přispělo 96 vývojářů, z toho 31 nových. Přehled novinek v příspěvku na blogu GitHubu a v poznámkách k vydání.

Vědci z Googlu vytvořili model pro předpovídání počasí a klimatu, který kombinuje umělou inteligenci a fyzikální simulace. Tento model prý dokáže dosáhnout špičkových výsledků, přičemž vyžaduje výrazně nižší výpočetní výkon než srovnatelné meteorologické modely provozované na ...

Cybersecurity company Acronis is warning that a now-patched critical security flaw impacting its Cyber Infrastructure (ACI) product has been exploited in the wild. The vulnerability, tracked as CVE-2023-45249 (CVSS score: 9.8), concerns a case of remote code execution that stems from the use of default passwords. The flaw impacts the following versions of Acronis Cyber Infrastructure (ACI) - &

Cybersecurity company Acronis is warning that a now-patched critical security flaw impacting its Cyber Infrastructure (ACI) product has been exploited in the wild. The vulnerability, tracked as CVE-2023-45249 (CVSS score: 9.8), concerns a case of remote code execution that stems from the use of default passwords. The flaw impacts the following versions of Acronis Cyber Infrastructure (ACI) - &Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

Anglická Wikipedie je k dispozici už také v černém. Na desktopu si lze vybrat světlý, tmavý nebo automatický vzhled.

Značka HMD Global se chce blíže spojit s rodiči dětí. • Chce vyvinout a začít prodávat telefony pro děti... • ...ideálně bez podpory sociálních sítí. Povede se to?

Several AI companies have recently been accused of collecting data used to train large language models (LLMs) without the consent of the affected parties.

The latest of the accusations come from Ifixit and Freelancer, which say Anthropic has collected data from the sites, even though they used a protocol to prevent that from happening. According to Freelancer CEO Matt Barrie, their site received 3.5 million hits from Anthropic’s Claudebot in four hours, making the bot “the most aggressive” to date.

Ifixit CEO Kyle Wiens, in a comment quoted by Engadget, said Anthropic is taking content without paying for it and forcing iFixit to use its own developer resources to fend off the data collection scheme.

Apple will reportedly delay introduction of Apple Intelligence on iPhones until it ships the iOS 18.1 update after the iPhone 16 ships this fall. I don’t see this as solely because the tech won’t be ready — developers are already testing it — but suspect regulatory challenges and Apple’s own wider deployment plans led to the delay.

The Artificial Intelligence (AI) hype is all encompassing. The industry is spending billions on it, electricity grids are struggling to maintain it, and regulators are preparing to constrain it. It is a speeding train packed with potential, but momentum is so rapid a mistake could send it off the rails. 

That’s a concern across the industry, one regulators are also attempting to understand. US regulation seems voluntary right now, while in Europe the much tougher EU Artificial Intelligence Act should come into effect this year. 

Apple joins the White House 

In the US, Apple has joined the White House Voluntary AI Safeguards program with 15 other major firms, including Amazon, Google, Meta, and Open AI. The aim of the group is to move toward safe, secure, and transparent development of AI technology.

The goal: to “mitigate AI’s safety and security risks, protect Americans’ privacy, advance equity and civil rights, stand up for consumers and workers, promote innovation and competition, advance American leadership around the world, and more,” the White House said.

While there is always a big element of poacher-turn-gatekeeper in any voluntary industry group, this slightly more laissez-faire approach will probably benefit the industry.

Europe is tougher

It’s different in Europe, where the Act takes the form of a sprawling piece of legislation that will take time to fully comprehend and implement. I expect the complexity of this law means most providers in the AI space will eventually mimic Apple and delay the introduction of services while they figure out how to be compliant. The laws are also being introduced in a staggered way across three years, which could make it harder to reach compliance.

Similar laws are being put in place globally, creating a complex regulatory environment in which most AI services will be forced to slow new product integration. For Apple as a platform provider, the regulatory complexity is amplified.

In this context, it makes a lot more sense for the company to switch on any of Apple’s new Intelligence services only once it has achieved enough clarity to guarantee compliance, particularly in the EU where the company has said it’s delaying the services pending such clarity. (Outgoing EU Commissioner Margrethe Vestager’s seemingly antagonistic response to that request underlines why Apple was concerned.)

Smoke and fire

That’s not to say there’s no smoke at all around this potential fire. Apple Intelligence will not be an iPhone-only animal, it will be available across the company’s entire ecosystem: iPhones, iPads, and Macs. That means the system needs to be widely tested across all these products, including analysis around compliance (above).

We know Apple is likely to have news to share about Macs and iPads this fall, as that’s when the company usually updates its hardware. That news is likely to include the introduction of new Apple Silicon processors, and it’s almost a certainty it will lean deep into its core messages around privacy, edge device AI, secure AI, energy and hardware integration when it does introduce new hardware.

Happy Thanksgiving

Teasing out those launches with the introduction of new AI features across its operating systems will only boost attention around the launch of new hardware. That attention should turn into sales, particularly as we hit the US shopping season and computer users consider the personal and economic consequences of the recent Microsoft/Crowdstrike failure. Against this backdrop, there’s never been a better time to introduce the world’s most advanced and best-designed hardware equipped with the world’s safest and most privacy-conscious form of AI service, Apple Intelligence. 

Summing up

With all of this in mind, I find it hard to be too concerned about Apple’s “delay” in launching its AI service. Regulation and its own internal product launch plans mean a later launch will still excite consumers, while helping it realize the much anticipated bounce in hardware sales everyone now expects as AI goes mainstream. I’m just not entirely certain any of us are truly ready for what the consequences of mass market AI might be. 

Please follow me on Mastodon, or join me in the AppleHolic’s bar & grill and Apple Discussions groups on MeWe.