Blockchain Council’s Post

Here are 4 steps to help spot AI manipulation.

In case you missed it, the Hugging Face team has published a great blog post on the 101 of #watermarking. It is especially pertinent in our current digital landscape, which is being increasingly populated with ‘deepfakes’ and AI-generated content. It explains the methods of watermarking AI-generated imagery, textual content, and audio, discussing each modality’s specific nuances. The article also reviews the pros and cons of different approaches and introduces the tools available on the Hugging Face Hub for adding and detecting watermarks I particularly loved the post’s clear structure and the inclusion of relevant examples. link: https://lnkd.in/gPBWjfcy #Watermarking #Deepfakes #AI #MachineLearning #HuggingFace

4chan users are leveraging AI tools to unleash a flood of racist images online. This disturbing trend highlights the dark side of AI and its potential for misuse. How can we combat the spread of harmful content while still harnessing the power of AI for positive change? #AI #Policy #Racism

Welcome to the era of (ir)responsible AI. Google recently paused the image generation of people in Gemini, its Large Language model, after the model became "way more cautious than we intended and refused to answer certain prompts entirely." The key phrase here is "became way more cautious than we intended." Is this the start of runaway-AI rather than irresponsible AI? #AI #Google #Gemini #ResponsibleAI #RunawayAI

If you don’t know where to start with your AI research journey, save this post now. I’ve been neck-deep in AI for the last 18 months or so, and it’s been a LOT of experimenting and researching. I honestly feel like YouTube owes me money for all the content I have watched. So, to save you some time, I've put together an AI starter pack that's got a bit of everything. This isn't the be-all and end-all, but it's a damn good place to kick things off… YouTube Channels to Watch ↳ Matt Wolfe: Consistent, energetic, and all about creative AI. Your weekly digest of all things AI. It's like a cheat sheet, but fun. https://lnkd.in/eK-AkHgp ↳ David Shapiro: Deep diver into societal impacts of AI. Macro views, big thoughts, and Star Trek uniforms. https://lnkd.in/ehATRQ8W ↳ Matthew Berman: Jack of all trades on AI. Whether you want the big picture or a detailed analysis, Matthew's your man. https://lnkd.in/e-dvHKRP  ↳ AI Explained: The voice of reason in the AI buzz. Academic with a dash of opinion zero hype. https://lnkd.in/e7ADf29A Websites to Explore ↳ There's An AI For That: Find your AI match here, no matter what you need. https://lnkd.in/e5RUkQvb ↳ AI Weirdness: Janelle Shane's playground of AI creativity. Get inspired, have a laugh. https://lnkd.in/eHHGfiA6 ↳ AI Dungeon: Storytelling meets gaming. Unique way to engage with AI and have fun while at it. https://aidungeon.com/ Especially Creative AI Tools ↳ Sora: Built by OpenAI, watch text prompts transform into full-fledged videos. Prepare to have your mind blown. https://openai.com/sora ↳ HeyGen: An avatar that reads scripts in your voice?! https://www.heygen.com/ ↳ Adobe Photoshop Firefly: Retouching made easy with Generative Fill. https://lnkd.in/ebPt8YP8 ↳ LTX Studio: A powerful video content creation tool to watch out for. https://ltx.studio/ ↳ Pi AI: Not your typical creative tool, but a fascinating conversational, voice-activated platform for some AI fun. https://pi.ai/talk ↳ Gemma.ai: Of course, our very own Gemma. Creativity and AI, all in one place: https://lnkd.in/ePhKcX6p Like I said, very much the tip of the AI iceberg, so feel free to drop in your recommendations in the comments! Don't hold back!

Ultimate Guide to AI Deepfake Technology A deepfake is a type of synthetic media where the likeness of someone in an existing image or video is replaced with someone else’s likeness using artificial intelligence. This technology utilizes sophisticated #AI algorithms to create or manipulate audio and video content with a high degree of realism. #Deepfake technology represents one of the most intriguing – and controversial – advancements in artificial intelligence today. As deepfakes become more accessible and their quality continues to improve, they pose significant challenges and opportunities across various sectors, including media, entertainment, and security. Understanding Deepfakes Creators of deepfakes use software that leverages advanced artificial intelligence and deep learning techniques, generative adversarial networks (GANs) in particular, to create or alter video and audio content convincingly. If you’ve been spending a considerable amount of time online over the past few years, you’ll have noticed that this technology started becoming popular in the contexts of celebrity videos and politics. But it’s now expanded its reach across various sectors, including entertainment, politics, and education. Deepfakes are created by training AI systems with large datasets of images, videos, or audio clips to recognize patterns in facial movements, voice modulation, and other physical attributes. The AI then applies this learned data to new content, superimposing one person’s likeness onto another person’s with striking realism. While the technology showcases remarkable advancements in AI, it also raises significant ethical and security concerns due to its potential for misuse, such as creating false representations of individuals saying or doing things they never did. 3 Positive Deepfake Examples. - Filmmaking and Entertainment - Education and Training - Healthcare and Therapy 3 Malicious Deepfake Examples -Political Misinformation - Financial Fraud - Image Abuse Bottom Line: Navigating Deepfake Technology Deepfake technology presents transformational innovative possibilities, but it’s often overshadowed by the challenges it presents. There’s no denying the immense potential of deepfakes to enhance areas such as entertainment, education, and even healthcare. However, the risks associated with their misuse in spreading misinformation and violating personal privacy cannot be overlooked. To ensure that the development and application of deepfake technology benefit society as a whole, we need to champion an approach involving innovation, rigorous ethical standards, and proactive regulatory measures. Full Article : https://lnkd.in/g-bhCjHw Helen E. Ijewere #deepfake #deepfaketechnology #cybersecurity #cyberrisk #dataprivacy

The fascinating thing in engaging with AI evangelists is understanding how little they know about the creative processes that they hope that AI will take over. There has been a curious but misguided push to have generative AI envelop, and subsume the creative arts. And the more I watch and chronicle the more hilariously, amusing and misguided it appears. The narrative that is pushed by AI enthusiasts is that creatives are afraid. No, we’re actually kind of amused. We’re not actually afraid. We’re starting to think of the whole thing as farcical, as we watch the surface of the bubble glisten and stretch before it shatters. Nonsensical AI enthusiast are saying AI is inevitable not realizing that AI has been used as essential technology in games for 4 decades and in film for 2.5 decades. Like the metaverse bubble before, in the generative AI bubble, we are seeing bandwagon riders, and bubble jumpers who want to forecast that their new favorite thing is going to take over, that it is inevitable, that it will reshape markets, and that it is doom. They are amusing. Why is the current generative AI movement so amusing? It is because they are bringing the wrong tools to the solution. And their tools are not as extensible as they think. Most importantly, they know nothing about the process that they hope to replace. Simply they screwed up when they came for large scale team development. That is the focus of games, animation, film, & TV. We got this. Taking games as an example. You don’t know how or why the game was built if you look at some video frames of that game out of context. Good games are fractions of reality boiled down to core ideas and then built up into systems by interdisciplinary teams that make magic through the chemistry of the people who game develop together. Here’s a simple analogy for the AI enthusiast who won’t understand what I just said. If you give someone a photo of a human, they possess a superficial understanding of that person as a snapshot in time on the surface. If an alien looks at that picture, they won’t know the underlying biology nor the systems that run inside of that human. And they cannot diagnose why that human is sick from one photo selected randomly across the lifetime of that human. Nor can they have any idea what systems run inside of a human if they do not know human biology. The alien is the LLM, without intelligence, without deduction. A film is more than a sequential series of images. To generative AI, a film is only a series of sequential images. But to human society, a film is never just a sequential series of images. To filmmakers, they know that the sequential series of images that they put out not only is completely and utterly unlike what an LLM will output, but the LLM can never understand the years long team process of creative development they went through. Simulation, iteration, persistent memory, & human learning are required for creation. Deal with it. #ai #fakeai #genai #degenai #aifail

Have you ever encountered a deepfake? To be honest, I probably have and didn't even realize it with how powerful technology is getting. And I'm not the only one... Deepfake scams involving Tom Hanks, Gayle King, and even MrBeast were a hot topic in the world of AI last week. Read the full story in Quad's latest #AI roundup below.

Text-to-image AI models can be tricked into generating disturbing images https://lnkd.in/gHf9Vsey AI News, AI, AI tools, Artificial intelligence – MIT Technology Review, Innovation, itinai.com, LLM, Rhiannon Williams, t.me/itinai 🚨 Text-to-Image AI Models Can Generate Disturbing Images: A Wake-Up Call for Better AI Safety 🚨 Researchers have found that popular text-to-image AI models can bypass safety filters, resulting in the creation of inappropriate and harmful images. This discovery highlights the urgent need for stronger security measures in AI systems. 🔐 SneakyPrompt: How It Works 🔐 A method called "SneakyPrompt" has been developed to exploit AI models and generate banned images. By using reinforcement learning, SneakyPrompt creates prompts that appear as gibberish to humans but are interpreted by AI models as requests for explicit content. This allows for the easy generation of inappropriate images. ❌ Existing Safety Filters Are Insufficient ❌ Although major text-to-image AI models like Stability AI's Stable Diffusion and OpenAI's DALL-E 2 have safety filters in place, SneakyPrompt can bypass these filters, exposing the ineffectiveness of the current guardrails. 🔒 Addressing the Issue 🔒 AI companies must develop stronger safety filters to protect against vulnerabilities like SneakyPrompt. One potential solution is to assess the tokens within prompts rather than the entire sentence to catch inappropriate requests. Additionally, blocking prompts containing words not found in dictionaries may help, although nonsensical combinations of standard words can still be used to generate explicit content. 💡 Practical Solutions for AI Implementation 💡 To leverage AI and stay competitive, follow these steps: 1️⃣ Identify Automation Opportunities: Find customer interaction points that can benefit from AI. 2️⃣ Define KPIs: Ensure your AI initiatives have measurable impacts on business outcomes. 3️⃣ Select an AI Solution: Choose tools that align with your needs and allow customization. 4️⃣ Implement Gradually: Start with a pilot, gather data, and strategically expand AI usage. 📚 Learn More and Get in Touch 📚 To learn more about AI KPI management and how it can redefine your business, contact us at [email protected]. For continuous AI insights, follow us on Telegram at t.me/itinainews or Twitter @itinaicom. 🌟 Spotlight on a Practical AI Solution 🌟 Discover the AI Sales Bot at itinai.com/aisalesbot. This solution automates customer engagement 24/7 and manages interactions across all stages of the customer journey. Redefine your sales processes and improve customer engagement with AI. Explore solutions at itinai.com. 🔗 List of Useful Links 🔗 🔬 AI Lab in Telegram @aiscrumbot – free consultation 📰 Text-to-image AI models can be tricked into generating disturbing images 🤖 Artificial intelligence – MIT Technology Review 🐦 Twitter – @itinaicom

I applaud the way Google is handling the recent Gemini Image Generation issues. It's an excellent example for transparency around AI. "So what went wrong? In short, two things. First, our tuning to ensure that Gemini showed a range of people failed to account for cases that should clearly not show a range. And second, over time, the model became way more cautious than we intended and refused to answer certain prompts entirely — wrongly interpreting some very anodyne prompts as sensitive. These two things led the model to overcompensate in some cases, and be over-conservative in others, leading to images that were embarrassing and wrong." #google #gemini #ai #transparency https://lnkd.in/gAC8YvQ6