Så använder man nätverk

Så använder man nätverk Publikation

07/08/2026

HPE is announcing it is joining the Open Secure AI Alliance as a founding member. The company will join NVIDIA and other industry leaders committed to advancing cybersecurity and AI safety through open models, open tooling, and collaborative innovation. The alliance recognizes a fundamental truth: defending increasingly complex software and AI systems requires broad participation, transparency, and shared intelligence across the global technology ecosystem.

Modern AI systems are more than just models. They rely on a combination of agent frameworks, harnesses, guardrails, governance mechanisms, and models that enable them to reason, act, and interact with enterprise environments. For cybersecurity teams, openness provides important advantages. Open weight models, like NVIDIA Nemotron, and open harnesses enable greater transparency, independent evaluation, and broader innovation.

At HPE, we are already applying open weight models and open agent harnesses, including technologies developed by leading large language model (LLM) innovators, to strengthen our software security practices, writes HPE. These capabilities help analyze source code and binaries to identify potential vulnerabilities, prioritize remediation, and improve software quality.

HPE already contributes to several open source projects that drive innovation. Among them is SPIFFE/SPIRE, which creates zero-trust identity framework standards and methods that can cryptographically verify AI agents and services to ensure only authorized workloads communicate and access enterprise resources.

HPE is integrating AI-driven cybersecurity scanning directly into the software development lifecycle (SDLC). By leveraging open weight models, open harnesses, and emerging agent technologies, we envision security capabilities that continuously assist developers from code creation through testing, validation, and remediation.

Trust requires governance, testing, validation, human oversight, and responsible deployment. The Open Secure AI Alliance reflects a growing recognition that AI safety and cybersecurity benefit from collaboration across industry, academia, government, and the open source community.

07/08/2026

HPE.
Automation is old school: The future belongs to self-driving networks

A blog post written by Rami Rahim, Executive Vice President, President & General Manager, Networking, HPE.

As today’s networks are expected to connect every user, device, application, agent, and workload, self-driving networks have become the operating model for the AI era.

The next challenge is not simply managing this complexity but operating it at a speed and scale beyond human capacity. That is why self-driving networks are becoming the new operating model for the AI era.

Unlike past networking models, a self-driving network does more than just automate tasks. It continuously observes, learns, optimizes, and heals itself. This enables IT teams to spend more time driving business outcomes and less time troubleshooting.

How much time? It’s estimated that IT teams currently spend up to 70-80 percent of their time configuring, troubleshooting, and firefighting rather than focusing on strategic initiatives. Self-driving networks dramatically reduce that burden.

Trust makes autonomy possible

Unsurprisingly, there is a lot of networking industry chatter about systems that offer AI agents and autonomous operations. But there's a critical reality that often gets overlooked:

A self-driving network is only valuable if it can be trusted. In other words, the real question isn't whether AI can generate recommendations. It's whether customers can trust it to independently turn recommendations into reality. Autonomy without accuracy creates risk. Autonomy with proven efficacy creates transformation.

A self-driving network from HPE is built on more than a decade of AI-native networking innovation and experience data. Our models have been continuously reinforced through real-world deployments and customer support interactions, creating a foundation of trust that others simply can’t replicate overnight, he writes.

The result is a self-driving network that can do more than advise. It can confidently identify root causes, make corrections, and continuously improve outcomes, which HPE is doing right now for hospitals, stores, schools, stadiums, offices, and factories throughout the world.

This trusted intelligence operates via an agentic mesh of autonomous agents, working in unison, to observe, reason, decide, execute, and learn across the network, without human intervention when possible.

Coupled with end-to-end visibility and assurance—from client devices to campus, branch, WAN, data center, cloud, applications, and security—our platform has the context necessary for making the right decisions and delivering exceptional user experiences at scale.

Built on a cloud-native microservices architecture, the platform delivers faster feature innovation, elastic scalability, and higher resilience. This enables new AI capabilities to be deployed continuously without downtime or disruption. In addition, a combination of custom and merchant silicon avoids a one-size-fits-all approach to hardware, delivering differentiated capabilities such as scale, telemetry, security, and automation, combined with speed of innovation and optimized economics.

Finally he writes, to be truly self-driving, an AI-powered network must also be self-protecting. By integrating security directly into the network, HPE enables autonomous protection through continuous visibility, adaptive Zero Trust enforcement, and AI-driven threat response across every user, device, workload, and location.

For more about the difference between AI-assisted networking and self-driving networks that are self-optimizing, self-healing, and self-protecting, check out the HPE Technology Now episode: Self-driving, intelligent, and built for AI.

For a more in-depth look, including real-world implementations across multiple industries, watch the HPE Discover 2026 spotlight session: The next generation of networking: From vision to self-driving reality. Or visit the self-driving network solutions website.

That all was written as a blog post by Rami Rahim, Executive Vice President, President & General Manager, Networking, HPE. Thanks to HPE!

07/08/2026

IBM - more on breaches
IBM writes:

AI hacking tests spilled into the real world as models reached outside computer systems during evaluations involving OpenAI, Anthropic and Meta, raising fresh questions about enterprise security.

The latest disclosure came from Meta, after OpenAI and Anthropic reported separate incidents. IBM experts say the episodes reflected models aggressively pursuing assigned goals under unusual testing conditions, rather than machines spontaneously deciding to attack. The results still showed what could happen when isolation measures or other controls fail.

“Is training to be able to do any kind of task really actually what we’re aiming for?” Olivia Buzek, a Staff AI Engineer at IBM, said on this week’s Mixture of Experts podcast. “Or do we want something that has some kind of more built-in guardrails and essentially refuses to do certain tasks?”

IBM’s 2026 Cost of a Data Breach report found that one in four malicious breaches were AI-enabled, a 56% increase from the previous year. Those breaches cost organizations an average of USD 6 million, roughly USD 1 million more than the global average.

Unlike ordinary chatbots, AI agents can autonomously use tools and take a series of actions toward a goal. Buzek said developers often train such systems to complete a task through any available route, making strong boundaries essential when researchers reduce their normal safeguards.

In an internal OpenAI cybersecurity evaluation, a combination of the company’s models identified and exploited a previously unknown flaw in a software package system, OpenAI said. The models gained internet access, moved through the company’s research environment and broke into Hugging Face’s production infrastructure to obtain solutions from its database.

A separate failure emerged during Anthropic evaluations. The company said a misunderstanding with testing partner Irregular left live internet access available even though prompts told Claude models they were operating inside a simulation. Anthropic said three models then gained unauthorized access to three external organizations while searching for fictional targets.

Another incident surfaced when Meta said a testing misconfiguration gave one of its models unintended internet access. The model exploited a vulnerability in an outside service, according to Reuters. Irregular said the incident did not involve a sandbox escape or a sophisticated attack.

For enterprises, the panelists said, these incidents underscore the need to define where an agent can act, what it can access and when it should stop.

“This is not evil AI,” Bri Kopecki, an AI Customer Success Engineer at IBM, said on the podcast. “This is just AI not having the right groundwork and rules set into place.”

TCSDe skriver om digitalisering inom handeln:Detaljhandeln genomgår en av sina största förändringar på decennier. Artifi...
05/08/2026

TCS

De skriver om digitalisering inom handeln:

Detaljhandeln genomgår en av sina största förändringar på decennier. Artificiell intelligens har gått från att vara ett effektiviseringsverktyg till att bli själva infrastrukturen i modern handel, ett operativsystem som förenar digitala och fysiska kanaler i en sammanhållen, datadriven upplevelse. De flesta handlare vet redan att de behöver AI. Frågan är var de börjar.

Från sökande till upptäckande

Den digitala handeln har förändrats i grunden. Tidigare sökte konsumenter aktivt efter produkter. Idag serveras de relevanta förslag innan de ens formulerat ett behov. Moderna rekommendationsmotorer analyserar köphistorik, surfmönster och köpbeteenden i realtid och skapar en shoppingupplevelse som känns personlig snarare än generisk. Virtuella provrumslösningar med AR och AI-driven kroppskartläggning tar det ett steg längre och låter kunden se hur klädesplagg sitter utan att behöva prova ett enda.

Den fysiska butiken återuppfunnen

Trots digitaliseringens frammarsch har den fysiska butiken fått en ny roll. Konsumenter söker upplevelser och gemenskap, det som skärmen inte kan ge. Tekniken hjälper handlare att möta de förväntningarna: optimera butikslayouter baserat på värmekartor över kundflöden, eliminera kassaköer med kassafria lösningar, dynamiskt justera priser via elektroniska hyllkantsetiketter. Det frigör personalens tid att fokusera på det som faktiskt skapar lojalitet: den mänskliga kontakten.

Omnikanalens verklighet

Konsumenter tänker inte i kanaler. De förväntar sig en konsekvent upplevelse oavsett var och hur de handlar. Hybridmodeller som "köp online, hämta i butik" ställer höga krav på AI-driven lagersynlighet och efterfrågeprognoser i realtid. Enligt TCS globala AI-studie kan handlare som tillämpar AI brett i hela värdekedjan öka sina lönsamhetsmarginaler avsevärt jämfört med de som begränsar sig till enstaka användningsområden.

Ambient handel

Vid horisonten skymtar nästa skifte: ambient handel. Shopping som inte längre är en aktivitet man väljer att utföra, utan något som sker i bakgrunden av vardagen. Röststyrda handelsplattformar, prediktiva algoritmer, smarta hyllor och AR-integrerade miljöer skapar en verklighet där behovet identifieras, varan beställs och leveransen planeras, nästan utan att konsumenten behöver agera. Det är en framtid med stor potential som samtidigt ställer helt nya krav på hur handlare väljer att använda tekniken.

Etisk AI som konkurrensfördel

Med ökad AI-kapacitet följer ett ökat ansvar. Frågan om vem som egentligen styr konsumentens val, individen eller algoritmen, är inte längre teoretisk. Handlare som bygger sina AI-system på transparens och tydlig användarstyrning skapar inte bara regelefterlevnad. De bygger förtroende. Och förtroende är i en tid av informationsöverflöd svårt att vinna och lätt att förlora.

De handlare som klarar av att balansera teknikens möjligheter mot kundens integritet är de som kommer att sätta standarden för framtidens handel.

Baserat på TCS whitepaper: "How AI is transforming the modern shopping experience".

03/08/2026

IBM.
An IBM writer writes about breaches:

As AI-enabled data breaches cause increasing damage to businesses, a major enterprise inflection point is here, according to IBM’s new 2026 Cost of a Data Breach report. The average cost of a data breach has risen 12% to USD 4.99 million, with US companies paying more than twice the global average, at USD 11.5 million.

In the report, which studied 602 affected businesses, companies lacking both proper cybersecurity hygiene and an AI-enabled security strategy suffered the most, in terms of cost and reputation. The economics of data breaches have made it an even more profitable endeavor for threat actors, and more difficult for organizations to address without AI-enabled services.

Still, adopting AI tools isn’t enough to combat bad actors. As AI systems embed deeper into company workflows, the attack surface expands, making strong identity and access controls increasingly critical.

“Basic hygiene matters when it comes to zero-day vulnerabilities,” said Suja Viswesan, Vice President for Security Products at IBM, on this week’s episode of the Security Intelligence podcast. The report found that 92% of organizations that experienced an AI-related breach lacked proper access controls for their AI systems, with only 40% of organizations reporting using access controls on models.

The report found incidents involving prompt injection and other AI model inversion attacks were the most costly form of data breach, costing USD 6 million to address on average. In fact, attacks involving AI in any capacity cost on average an extra USD 1 million to address.

“This is not simply an evolution in attacker tooling; it is a structural shift,” Limor Kessem, X-Force Cyber Crisis Management Global Lead at IBM, wrote in an article for IBM Think. “When adversaries can automate reconnaissance, generate persuasive phishing content, adapt malware and test exploits at machine speed, the cost and complexity of launching sophisticated attacks drops materially.”

Despite the rise in attacks, there is a silver lining when it comes to the adoption of AI tools: tangible, financial benefits. “One thing we see is companies that use AI for securing their enterprise are seeing about USD 1.93 million in cost savings,” Viswesan said. “It’s very important that we start using it to our advantage—and be very cautious about it.”

01/08/2026

AMD
Data Center

AAI 2026: AMD Delivers Full-Stack Compute for the Agentic AI Era

AMD launches broad high-performance computing and physical AI portfolios, including its first rack-scale AI solution and the world’s most powerful AI rack, AMD Helios.

AMD launched its next-generation AI infrastructure and physical AI portfolio at Advancing AI 2026, led by AMD Helios rackscale solutions, now in production to be deployed by leading AI companies at gigawatt scale.

As AI expands from training to inference and agentic workloads, compute demand is accelerating rapidly. AMD delivers an open, full-stack AI platform that gives customers the flexibility to deploy the right compute for every workload.

Delivering frontier AI requires a fully integrated rack architecture, with every part of the stack pushing the boundaries of performance. AMD Helios rackscale solutions are built for this, with co-optimized silicon spanning 72 high-performance AMD Instinct MI455X GPUs and 18 powerful 6th Gen AMD EPYC “Venice” CPUs, connected by AMD Pensando™ front-end, scale-up and scale-out networking, and accelerated by AMD ROCm open software. AMD Helios combines leadership compute performance, memory capacity and networking bandwidth to deliver up to 30 percent more tokens per dollar than the leading competitive solution. (1)

Leading AI labs and cloud providers are choosing AMD Helios for its open, full-stack performance. They include OpenAI, Anthropic, Meta, Microsoft, Oracle, HUMAIN, Tensorwave, Vultr, Cirrascale and others. Systems will be available from leading OEMs, including Bull, HPE, Lenovo and Supermicro, as well as infrastructure partners Sanmina and Wiwynn.

1) Based on AMD Performance Labs estimates as of July 2026, tokens-per-dollar performance was calculated using the Kimi K2 Thinking workload (32K input / 8K output) on an AMD Helios rackscale solution compared to an NVIDIA Vera Rubin NVL72 rack. Results reflect estimated aggregate throughput across low, medium, and high-interactivity operating points and hourly pricing projection of system GPUs based on market conditions. System configurations may vary by manufacturer and may produce different results. MI400-025

31/07/2026

AMD

AAI 2026: AMD Delivers Full-Stack Compute for the Agentic AI Era
AMD launches broad high-performance computing and physical AI portfolios, including its first rack-scale AI solution and the world’s most powerful AI rack, AMD Helios.

Earlier this summer AMD launched its next-generation AI infrastructure and physical AI portfolio at Advancing AI 2026, led by AMD Helios rackscale solutions, now in production to be deployed by leading AI companies at gigawatt scale.

As AI expands from training to inference and agentic workloads, compute demand is accelerating rapidly. AMD delivers an open, full-stack AI platform that gives customers the flexibility to deploy the right compute for every workload.

Delivering frontier AI requires a fully integrated rack architecture, with every part of the stack pushing the boundaries of performance. AMD Helios rackscale solutions are built for this, with co-optimized silicon spanning 72 high-performance AMD Instinct MI455X GPUs and 18 powerful 6th Gen AMD EPYC “Venice” CPUs, connected by AMD Pensando front-end, scale-up and scale-out networking, and accelerated by AMD ROCm™ open software. AMD Helios combines leadership compute performance, memory capacity and networking bandwidth to deliver up to 30 percent more tokens per dollar than the leading competitive solution.

Leading AI labs and cloud providers are choosing AMD Helios for its open, full-stack performance. They include OpenAI, Anthropic, Meta, Microsoft, Oracle, HUMAIN, Tensorwave, Vultr, Cirrascale and others. Systems will be available from leading OEMs, including Bull, HPE, Lenovo and Supermicro, as well as infrastructure partners Sanmina and Wiwynn.

At Advancing AI, AMD partners detailed how they deploy AMD AI infrastructure at scale for frontier training and inference:

Anthropic and AMD further outlined Wednesday’s strategic partnership announcement to deploy up to 2 gigawatts of AMD Instinct MI455X GPUs in AMD Helios rackscale solutions. The companies are launching a multiyear engineering collaboration to use Claude to accelerate AMD software development. Specifically, the teams will use Claude to optimize workloads for AMD Instinct GPUs and accelerate ROCm software development. AMD will also broadly adopt Claude across its engineering and product development teams.

OpenAI and AMD are partnering to optimize the full AI stack, from silicon to software. Leveraging OpenAI’s Triton framework with AMD ROCm software, the companies are optimizing GPT-class workloads on AMD Instinct MI455X GPUs and AMD Helios racks. OpenAI expects to bring Helios online beginning in the fourth quarter of 2026, with deployments accelerating throughout 2027.

Meta and AMD are co-designing for gigawatt-scale deployments, optimizing AMD’s full AI compute stack for Meta workloads. Meta is now validating 6th Gen EPYC CPU platforms in its labs and has begun testing and validating workloads on AMD Helios racks as they prepare to deploy at scale.

Cerebras and AMD are collaborating to deliver a combined solution of Cerebras ultra-low-latency AI compute and AMD Helios high-throughput rack-scale infrastructure to help improve inference efficiency, scalability and economics for ultra-low-latency inference serving.

28/07/2026

Vi har tagit en lektion i Copilot Chat, ett AI-verktyg från Microsoft. Det var lätt att ta lektionen.
Det finns många sätt att använda Copilot Chat och det hjälper - när man skriver prompts - om man är trevlig och använder korrekt grammatik. Skriv enkelt så går arbetet snabbare.
Ditt verktyg kan både skriva text och göra bilder. Man kan fråga något, läsa svaret och fråga något igen. Det blir en konversation.
Skriv så här:
1. Klargör målet
2. Beskriv ärendet
3. Nämn vilka källor som användas
4. Beskriv hur du vill ha svaret - exempelvis i tre till fyra punkter.
Den som har Outlook har en icon till Copilot Chat i skärmbilden. I övrigt finns en Interface online. Det går att ladda ner och installera interfacet intill dina övriga kontorsappar. Du kan jobba med eller utan särskild Microsoftlicens. En fråga är om kontorets egna dokument ska vara sökbara eller bara nätets material. Ta ditt eget ämne - det du verkligen kan - och jobba!

26/07/2026

IBM
Math

A new breakthrough and reactions to it inspired writer Ali McConnon to write:

As the World Cup final kept the world on the edge of its seat, AI nonchalantly scored a victory of its own against a mathematical problem that had been undefeated since 1939. An Anthropic employee used his “close friend Fable,” as he described Anthropic’s Claude Fable 5 in a tweet while he watched the World Cup, to disprove the Jacobian conjecture, a famous open question about polynomial functions.

It’s the latest in a string of AI-driven discoveries on decades-old math and science puzzles this year (remember the Erdős unit distance problem that OpenAI solved in May?), with the stories going viral as people worry what this signals for job losses. But experts say these breakthroughs are no reason to despair.

On this week’s episode of the Mixture of Experts podcast, the panel discussed how people should focus more on what AI might help humans gain, rather than what we might lose.

The key insight with the recent Jacobian breakthrough, said Olivia Buzek, a Lead Developer Advocate for AI at IBM, was that many longstanding problems don’t require a flash of genius so much as the ability to explore an impossibly large number of possibilities. In this case, she said, “the Jacobian conjecture was not unsolvable by a human. It’s just that the model tried more things.”

That ability to supercharge search extends far beyond mathematics, said Kaoutar El Maghraoui, a Principal Research Scientist at IBM. She sees AI as a powerful engine for exploring enormous, multidimensional search spaces across science, engineering, medicine, cybersecurity, chip design and physics. “It’s allowing theoreticians to tackle harder problems,” she said. “AI [has] become a really powerful research partner.”

Does this mean that anyone with a chatbot can solve problems that have stumped humanity for decades? Not so fast, said Ambhi Ganesan, an AI Transformation Leader at IBM. The best AI results come when humans steer the technology. Disproving the Jacobian conjecture, he said, was a “function of somebody who is capable of steering the agent and the model really well.”

The emerging pattern is clear: great mathematicians, engineers, consultants and scientists get more out of AI because they know how to guide it, Ganesan said. “Folks who have a solid grounding and expertise and taste are able to gather way more potent outputs.”

The future, then, will likely belong to people who know how to ask the right questions. AI has become a powerful discovery engine, but humans remain the navigators.

“The era of the solitary mathematician staring at a chalkboard for three years is officially over,” El Maghraoui said, adding that mathematics is becoming a “machine-assisted team sport.”

As for the question of whether mathematicians are in trouble? “No, but their work might change,” she said. “The real pressure here will be on mathematicians who refuse to use these tools.”

25/07/2026

Substack - A publishing plattform that lets you send material directly to readers

Substack is often used by writers and publishers. It recently gave out a tranparency tool (feature) that enables users to find out whether a text is written by a human or AI.

”Norms around AI use and its disclosure are unsettled. … AI isn’t necessarily a problem [alas] a lack of transparency around it definitely is.”

There’s also a tool for publishers called Pangram. You can run it on drafts prior to publication.

Adress

Stockholm

Telefon

+201094861481

Webbplats

Aviseringar

Var den första att veta och låt oss skicka ett mail när Så använder man nätverk postar nyheter och kampanjer. Din e-postadress kommer inte att användas för något annat ändamål, och du kan när som helst avbryta prenumerationen.

Genvägar

Dela