Michael Larabel

There's been a lot of heated discussion lately around the value and usability of SBOMs. I won’t go into that debate here, but I will say this: interoperability and the ability to seamlessly exchange BOMs and other machine-readable product-related artifacts between different systems and parties are key to broader adoption and usability. Enter the CycloneDX Transparency Exchange API (TEA) Standard, also known as project Koala. This initiative is designed to make that interoperability a reality. It supports the automated exchange of crucial supply chain artifacts like SBOMs, VEX documents, and attestations, so users can easily discover and consume transparency-related data for any product. Curious to learn more? Join us for our public webinar on the OWASP Transparency Exchange API on Monday, Nov 25th, at 17:00 CET. Register for free and help spread the word! 🔗 https://lnkd.in/dDk9kXP6 OWASP CycloneDX SBOM/xBOM Standard #SBOM #SPDX #CycloneDX #SoftwareSupplyChainSecurity

39

6 Comments

It's in print! Just received my complimentary copy of the ECC book from Manning Publications Co. which I had the pleasure of reviewing a few months ago. An excellent, detailed deep dive into the implementation of cryptographic primitives, with enough background on the maths of Elliptic Curves to keep matters interesting. #ellipticcurve #cryptography #book

13

2 Comments

While PCIe fails hard, when you are trying to get everything through the kernels, the new hardware requirements, will make smaller setups to fail horribly without proper RDMA technology at scale. Current NICs are very inefficient, in the transfer scale needed for AI. This calls for another reason to go beyond modern quantization capabilities. This is a problem that will remain persistent, as these new RoCE devices/SuperNics, will have to wait for their economies of scale till they are commodity hardware. NVidia might be coming from Cisco at this rate.... #AI #funding #angelinvesting #Aistartup https://lnkd.in/gUBsheVa

1

It's always a good idea to flash your Google Pixel Phone with a recent version of GrapheneOS and eliminate backdoors like this one: Every Google Pixel Phone Has a Verizon App that Doubles As a Backdoor https://lnkd.in/e6aYvcgM

3

Here we go again. There's another Foxit Reader crash detected by EXPMON Public (https://pub.expmon.com). Unlike the previous ones that were actually patched in the latest version (https://lnkd.in/g_wPzfkJ), this one does cause a crash on the latest Foxit Reader. However, my brief analysis has found that: 1. This is not a malicious exploit, just a real-world PDF sample happens to trigger a bug in Foxit Reader. 2. It seems to be a null pointer dereference crash (so probably not exploitable). Nonetheless, I've sent the sample to the vendor (Foxit Software), for safety reasons. Will share out the particular sample after the vendor finishes their investigation. As reported, recently EXPMON Public has detected some "zero-day" PDF crashes including one notable crafted/exploitable Adobe Reader crash (https://lnkd.in/ghQYYzXR), when it's just on its mission to analyze a (huge) public PDF sample set - WILD. EXPMON is designed and built for detecting any exploitation-related activities. Because application crash is a big and important indicator for potential exploitation, EXPMON carefully reports them. So, if you see Indicator related to application crash when using EXPMON, you should pay attention to that submission. #expmon #ExploitDetetcionInnovations #foxitreader #pdf

24

Sharing the video of my presentation at MPLS SD & AI Congress 2024 on AI/ML networking... Thanks to the previous speaker who finished his presentation 5 minutes early. I got an extra 5 minutes to finish my talk!... 😊 By the way, all these years, I disliked hearing my own voice and would never listen to any of my recorded speeches. It turns out that this is quite a common phenomenon and has a name for it: "voice confrontation." It took me a while to get over it—mainly by repeatedly listening to my recordings and focusing on the content rather than the sound of my voice or the filler words I use... Now, I can confidently post my videos on LinkedIn! Try this technique if you have similar inhibitions - it works! https://lnkd.in/geb_YqFx #AI #networking #MPLSWC24 Juniper Networks

241

14 Comments

What’s post-quantum cryptography (PQC) got to do with the future of security? In this newsletter, Phil Venables explains why we need to prepare for PQC.

6

Here is a video link to YorChips presentation if you missed it. We cover Why Chiplets for future SOC Limitations on current UCIe Patented Technology enabling low cost packaging Enabling interop and adaptability with FPGA & CPU Chiplets. https://lnkd.in/gDEV93vS #UCIe #chiplets #SOC #networking #AI #LLM

46

2 Comments

Here is a paper about Chinese event detection. It proposes a novel model named Label-Aware Heterogeneous Graph Attention Network (L-HGAT). https://lnkd.in/gE97gc-F #EventDetection #HGAT #LabelEmbedding

1

1 Comment

Interesting, expecially in light of the fact that AI is really only just beginning to make its way into regular contact with people outside the relatively small number of us who cover it closely. Once upon a time, when most software came in a box with a CD (or even earlier a floppy disk) in it, such a box had its requisite bullet points of features used for marketing purposes, usually with the latest buzzwords attached that the product absolutely needed to have. If AI-based apps were distributed boxed in the same way, we'd be entering the period where marketers are quietly scrubbing "AI Inside" from the list of bullet points. AI is not going away, but I've noticed that consumer-facing AI is just simply not catching fire. This is a good sign, I believe, because it means that companies are now being more selective about where and how they use the technology as they work with it more, and they are evaluating it increasingly in the same way that any new consumer technology is evaluated: "Does it improve the user experience?"

12

10 Comments

Western Digital has recently released it's newest Blue SN5000 NVMe SSD in a PCIe 4.0 design and, although there isn't much to scream and shout about, this SSD is a bit different from the rest.  It relies on two separate types of memory, depending on the chosen size.  If you are buying the SN5000 from 500GB to the 2TB capacity, the memory in use is KIOXIA 112-layer 3D TLC BiCS 5.  ##DMark #162layer #4TB #analysis #Anvil #ASSSD #ATTO #BiCS5 #BiCS6 #CrystalDiskInfo #CrystalDiskMark #DRAMLESS #KIOXIA #PCMark10 #performancetesting #QLC #report #Review #SN5000 #TxBench #WDBLUE #WesternDigital

6

This is a very technical video, but here Pablo Acosta-Serafini explains a common mistake with handling secrets (i.e. credentials) in Node-RED/NodeJS. Why do you need enterprise support from a team that has deep expertise? This is why:)

16

I recently sat down with Mike Vizard from TechStrong.ai to discuss how Phison's innovative aiDAPTIV+ technology is making generative AI accessible and affordable for small to medium-sized enterprises. We explored the barriers to entry and how our latest aiDAPTIV+ Pro Suite is simplifying the adoption of Gen AI. You can listen to our full conversation here: https://lnkd.in/ekBdzEk4 #Phison #AI #ML #LLM

27

I'm advocating for using markdown as a simple file format to recording and replaying LLM sessions, specifically using bird tracks before prompts. There are lots of ways to render markdown. HaverScript uses this format already for the echo mode. Thoughts? https://lnkd.in/gbfN7mBf

5

1 Comment

https://lnkd.in/gqp4xTXS Nvidia's Blackwell chip is delayed, the company tells big customers. Design flaw prevented the start of mass production. #nvidia #artificialintelligence #llms #chatgpt #generativeai #gpu #servers #datacenter #cloud #ai #technology #innovation #business #Semiconductorindustry #electronics #semiconductor #supplychain credit: the information

34

1 Comment