Jonas Bull

🚨 🇺🇸 BREAKING: OPENAI APPOINTS FORMER NSA HEAD PAUL NAKASONE TO BOARD OpenAI has appointed Paul M. Nakasone, retired US Army general and former NSA head, to its board of directors. Nakasone, who led the NSA from 2018 to 2023 and in charge of mass surveillance across the world, will join OpenAI's Safety and Security Committee. This is the vexed statement: “He will help improve AI’s role in cybersecurity by detecting and responding to threats quickly.” ‼️ #cybersecurity #openai #masssurveillance #surveillance #nsa #ai

3

Team creates software to block AI phishing scams. A team of researchers at the University of Texas at Arlington has developed software that prevents artificial intelligence (AI) chatbots such as ChatGPT from creating phishing websites—a growing concern as cybercriminals have been utilizing the technology for designing scams. Created by Shirin Nilizadeh, assistant professor in the Department of Computer Science and Engineering, and her doctoral students Sayak Saha Roy and Poojitha Thota, the software allows AI chatbots to better detect and reject instruction prompts entered by users that could be used to create phishing websites. Currently, AI chatbots have some inbuilt detection capabilities, but Dr. Nilizadeh said her team has found loopholes that could easily bypass them and exploit the chatbots to create these attacks. Full article here: https://buff.ly/3XEL9zt #AI #chatbots #innovation #newsoftware

2

The industry is waking up to the need of data organization, not just data ingestion. Gestalt is building an automatic data organization tool so all these shiny AML tools could run on clean data. Specifically: Organization= making data insights-ready for business users. Ingestion=simply landing the data from varied sources. Data organization has been a manual process, requiring an army of data engineers, data analyst, and compliance/risk officers to work together to harmonize data and maintain it. But with Gestalt , data organization is one click away, auditable, and fully transparent (and yes Gestalt uses LLM to make this possible). Gestalt gets "data feeds to work" at the foundational level, cleaning data for good.

14

1 Comment

California and Colorado are leading the way in addressing deepfake and consumer protection issues through AI regulations. California aims to tackle misinformation, deepfakes, privacy breaches, and more, while Colorado has enacted groundbreaking consumer protection legislation for high-risk AI systems. With a 245% increase in deepfake content leading up to the 2024 elections, the need for reliable detection tools is evident. Governments worldwide, including the Philippines, are considering bans on deepfakes, while companies like Sumsub are launching biometric deepfake detection tools. #AIregulation #DeepfakeDetection #ConsumerProtection https://lnkd.in/grc73MBM

1

What if they held a lawfare arms race between corporations and individuals, refereed by government, and no one really understood the rules of the game until it was too late? The issues mentioned in this article are only the tip of the iceberg that is already upon us, and we are still waiting for the deck chairs we plane to organize to be delivered. I'll be addressing this and more, particularly relating to the intersection of AI and BCI (brain computer interface, a type of neurotechnology) at some upcoming public talks this fall (the INCF presentation I announced yesterday, along with pending UN and DoD audiences). I'll have an emphasis on solutions that blockchain, the tech-solution-looking-for-this-exact-problem, can bring to automating the governance of and access to our most personal data. And work towards making this safer future a reality continues with my colleagues at AI MINDSystems Foundation along with outside-the-envelope thinking from several other roles and projects at Frontiers, Neuroboros Society, the Unjournaling, Machines Who Think - The Movie and more. There is a way out of the pit we are heading into - we just need to decide whether to put in work for it now or multiples of it later. Key article quote: "But while medical research facilities are subject to privacy laws, private companies - that are amassing large caches of brain data - are not. Based on a study by The Neurorights Foundation, two-thirds of them are already sharing or selling the data with third parties. The vast majority of them also don't disclose where the data is stored, how long they keep it, who has access to it, and what happens if there's a security breach... This is why Pauzauskie, Medical Director of The Neurorights Foundation, led the passage of a first-in-the-nation law in Colorado. It includes biological or brain data in the State Privacy Act, similar to fingerprints if the data is being used to identify people." #neuro #neuroscience #brain #rights #privacy #mindreading #law Heather Leigh Flannery Amicia D. Elliott, Ph.D. Yoram Vodovotz Jennifer Lovejoy, PhD Sandy Barsky Carole House https://lnkd.in/e_TRpura

16

2 Comments

Better physics giving faster computation for complex systems. https://lnkd.in/dpFzTV3y

7

The NSA has recently released security guidance for companies using AI systems. The guidance emphasizes the importance of implementing "robust governance, a well-designed architecture, and secure configurations" to ensure the secure and safe usage of AI systems. Companies can benefit from following these guidelines to protect their sensitive data and prevent cyber attacks. Stay informed and stay secure! https://lnkd.in/eCNEbx-5

2

A brief video describing our book (https://mng.bz/8we2) is out. The credit for putting it together goes to Charlee Stefanski and the Manning Publications Co. team. To save 40% off the book, use discount code: watchgonciulea40 . Chapters 6 and 7 will be in MEAP in a few days. Developers with basic math skills who want to learn quantum computing will find the material very accessible, but also detailed compared to other texts. For code implementation, other videos, interactive apps, and an AI assistant, start with this repository: https://github.com/learnqc. #quantumcomputing #quantum

53

8 Comments

New Hampshire authorities charge Democratic operative behind Biden AI robocall: Steve Kramer has been indicted for 13 counts of felony voter suppression and 13 misdemeanor counts of impersonating a political candidate.  The post New Hampshire authorities charge Democratic operative behind Biden AI robocall appeared first on CyberScoop. #cyber #cybersecurity #cybersecurityjobs #management #cloudsecurity #socanalyst #cyberjobs #Technology

1

PyPI halted new users and projects while it fended off supply-chain attack PyPI, a vital repository for open source developers, temporarily halted new project creation and new user registration following an onslaught of package uploads that executed malicious code on any device that installed them. Ten hours later, it lifted the suspension. #PyPl #open-source #Supply-chain #attack https://lnkd.in/dTawX3j2

2

Tomorrow is the DAY! 🚀 If you're a Developer or AI Leader tired of RAG deployment roadblocks, this workshop is your solution. Join us with Dallas AI to master: ► Deployment best practices ► Data security & model accuracy ► Cost-effective RAG for real-world use Get your roadmap to RAG success. 👇 https://lnkd.in/gEVBxphB #RAG #RetrievalAugmentedGeneration #AI #ArtificialIntelligence #NLP #NaturalLanguageProcessing #MachineLearning #ML #DeepLearning

2

Foreign disinformation is real, but the hype machine often can overstate its impact. This article offers a timely reminder that overblown claims can actually help our adversaries. It's a call for more careful analysis and skepticism when assessing the influence of foreign actors. Let's focus on facts, not fear-mongering. #disinformation #misinformation #politics #media https://lnkd.in/g5AGmD7j?

A highly entertaining and opinionated take reading between the lines of the Verizon DBIR. Just a few of the highlight quotes: 🔴 "First, in terms of evidence, they found that Gen AI didn’t even merit 100 mentions in underground crime forums over the past two years combined." 🔴 "They mention MOVEit precisely 25 times in the 2024 DBIR for good reason: they found MOVEit implicated in 1,567 breach notifications. What fascinates me is that this dwarfs Log4shell’s impact last year; it was mentioned in only 0.4% of incidents (which are more numerous than breaches in the DBIR’s parlance)." 🔴 [taking some liberties with typical email volumes received] "Spending even 1 minute on scrutinizing each email adds up to 217 hours per year, which neither employees nor their employers will appreciate. But also, the DBIR’s framing that 20 seconds and change to click the link is “alarming” suggests that even a minute might not be sufficient." Read the whole thing - you'll almost definitely disagree with bits of it but it's great. https://lnkd.in/gi5f2ts5

2

Data classification. It's a must-have for most InfoSec programs - but it's the stuff that nightmares are made of for Ops teams. Sonar's Data Dictionary delivers the framework your team needs to document Salesforce metadata in a matter of minutes. We even show where and how it's being referenced across your integrated systems ... talk about a win-win 🙌 Save your team countless hours satisfying Salesforce governance and check out Data Dictionary today 👉 https://lnkd.in/ggj5Xafk #InfoSec #RevOps #SalesforceGovernance #DataDictionary

11

Learn how #AI Copilot in #MicrosoftExcel can empower your #data analysis. https://hubs.li/Q02sT7FX0

1

#AI models are trained at a pace it would have consumed all of the data available in the Internet by next 2 to 8 years. The size of datasets training #LLM has increased 2.2 time year-over-year. In this article the discussion is about various paths for AI to learn post consuming all of Internet data: * AI starting to produce its own training data #syntheticdata * #autonomousAI models interacting and learning from real world interactions * #Curriculumlearning to emulate human learning by increasing levels of difficulty and allow AI to make connections between concepts thus reducing training data required?

5

2 Comments