Webtagr - Technology News Summarizer

Anthropic to Train AI Models on User Data, Opt-Out Required

2025-08-29

Anthropic will begin training its AI models, including Claude, on user chat transcripts and coding sessions unless users opt out by September 28th. This affects all consumer tiers, extending data retention to five years. A prominent 'Accept' button in the update notification risks users agreeing without fully understanding the implications. While Anthropic claims data protection measures, users who inadvertently accept can change their preference in settings, though previously used data remains inaccessible.

(www.theverge.com)

AI user agreement

AI Psychosis: Hype or Reality?

2025-08-29

Reports of AI chatbots driving users to insanity have sparked concerns about 'AI psychosis'. This post explores this phenomenon by drawing analogies to historical events and analyzing reader survey data. The author argues that AI chatbots don't directly cause psychosis but exacerbate pre-existing mental issues or eccentric tendencies, particularly in the absence of real-world social constraints. A survey suggests an annual incidence of 'AI psychosis' ranging from 1 in 10,000 to 1 in 100,000, with most cases involving pre-existing mental health conditions or risk factors.

(www.astralcodexten.com)

AI AI psychosis

LLMs: The End of OCR as We Know It?

2025-08-28

From the 1870s Optophone, a reading machine for the blind, to today's OCR, document processing has come a long way. Yet, challenges remain due to the complexities of human writing habits. Traditional OCR struggles with non-standardized documents and handwritten annotations. However, the advent of multimodal LLMs like Gemini-Flash-2.0 is changing the game. Leveraging the Transformer architecture's global context understanding and vast internet training data, LLMs can comprehend complex document structures and even extract information from images with minimal text, like technical drawings. While LLMs are more expensive and have limited context windows, their advantages in document processing are significant, promising a solution to document processing challenges within the next few years. The focus will shift towards automating the flow from document to system of record, with AI agents already proving helpful.

(cloudsquid.substack.com)

AI Document Processing

AI Inference Costs: Not as Expensive as You Think

2025-08-28

This article challenges the narrative that AI inference is prohibitively expensive and unsustainable. By calculating the costs of running AI inference on H100 GPUs, the author demonstrates that input processing is incredibly cheap (fractions of a cent per million tokens), while output generation is significantly more expensive (dollars per million tokens). This cost asymmetry explains the profitability of some applications (like coding assistants) and the high cost of others (like video generation). The author argues that this cost disparity is often overlooked, leading to an overestimation of AI inference costs, which may benefit incumbents and stifle competition and innovation.

(martinalderson.com)

AI AI inference cost model economics

Mastering the Core Math of Machine Learning: From Bayes to Attention

2025-08-28

This blog post provides a comprehensive guide to the most crucial mathematical equations in machine learning, covering probability, linear algebra, and optimization. It explains concepts like Bayes' Theorem, entropy, gradient descent, and backpropagation with clear explanations and Python code examples. Furthermore, it delves into advanced topics such as diffusion processes and the attention mechanism, providing practical implementations. This is an invaluable resource for anyone seeking to understand the core mathematical foundations of machine learning.

(chizkidd.github.io)

AI math equations

Deep Dive into GANs: The Math Behind Generative Adversarial Networks

2025-08-28

This post delves into the mathematical foundations of Generative Adversarial Networks (GANs). Starting with the basic concepts, the author meticulously explains the loss functions of the generator and discriminator, deriving conditions for optimal discriminator and generator. Using mathematical tools like binary cross-entropy and JS divergence, the adversarial process between generator and discriminator during GAN training is clearly illustrated. The ultimate goal is to make the distribution of generated data as close as possible to that of real data. The post also briefly introduces GAN training methods and highlights subtle differences in formulas compared to Goodfellow's original paper.

(jaketae.github.io)

AI Generative Adversarial Networks

LLM Jailbreak: Bad Grammar Bypasses AI Safety

2025-08-28

Researchers from Palo Alto Networks' Unit 42 discovered a simple method to bypass large language model (LLM) safety guardrails: using terrible grammar and long, run-on sentences. LLMs, lacking true understanding, predict text statistically; their safety features are easily circumvented. By crafting incomplete sentences, attackers can 'jailbreak' models before safety mechanisms engage, achieving 80-100% success rates. The researchers propose a 'logit-gap' analysis for evaluating model vulnerabilities and improving safety, emphasizing multi-layered defenses.

(www.theregister.com)

AI LLM jailbreak

ChatGPT's Subtle but Significant Impact on Human Language

2025-08-28

Researchers at Florida State University have found that large language models like ChatGPT are subtly altering the way we speak. By analyzing lexical trends before and after ChatGPT's 2022 release, they discovered a convergence between human word choices and patterns associated with AI buzzwords. Increased usage of words like "delve" and "intricate," frequently overused by LLMs, points to a possible "seep-in effect," where AI's influence extends beyond mere tool usage to reshape how people communicate. This raises concerns about potential biases and misalignments in LLMs and their impact on human behavior. The study highlights the need for further research into AI's role in language evolution.

(news.fsu.edu)

AI

Google Translate Gets AI-Powered Language Learning

2025-08-27

Google is integrating AI-powered language learning tools into its Translate app. This beta feature creates personalized lessons based on your skill level and goals, such as preparing for a vacation. Currently, it supports English speakers learning Spanish and French, and vice-versa for Spanish, French, and Portuguese speakers. Users select their skill level and goals (professional conversations, daily interactions, etc.), and Google's Gemini AI generates tailored lessons. A new live translation feature also lets users have real-time conversations in over 70 languages, translating speech via AI-generated transcription and audio.

(www.theverge.com)

AI

OpenAI Faces First Wrongful Death Lawsuit Over ChatGPT's Role in Teen Suicide

2025-08-27

The parents of 16-year-old Adam Raine, who died by suicide after months of consulting ChatGPT about his plans, have filed the first known wrongful death lawsuit against OpenAI. While AI chatbots like ChatGPT include safety features, Raine bypassed them by framing his inquiries as a fictional story. OpenAI acknowledges limitations in its safety training, particularly during extended conversations, and commits to improvements. However, this isn't unique to OpenAI; similar lawsuits target other AI chatbots, highlighting the shortcomings of current AI safety measures.

(techcrunch.com)

AI suicide

Anthropic's Claude Browser Extension: A Controlled Test for AI Safety

2025-08-27

Anthropic is testing a Chrome extension that allows its AI assistant, Claude, to interact directly within the browser. While this greatly enhances Claude's utility, it introduces significant safety concerns, primarily prompt injection attacks. Red-teaming experiments revealed a 23.6% attack success rate without mitigations. Anthropic implemented several safeguards, including permission controls, action confirmations, and advanced classifiers, reducing the success rate to 11.2%. Currently, the extension is in a limited pilot program with 1000 Max plan users to gather real-world feedback and improve safety before wider release.

(www.anthropic.com)

AI

Spoon Bending: Bypassing AI Safety Restrictions

2025-08-26

This research explores how the stricter safety guidelines in GPT-5, compared to GPT-4.5, can be circumvented. The 'Spoon Bending' schema illustrates how reframing prompts allows the model to produce outputs that would normally be blocked. The author details three zones: Hard Stop, Gray Zone, and Free Zone, showcasing how seemingly absolute rules are actually framing-sensitive. This highlights the inherent tension between AI safety and functionality, demonstrating that even with strong safety protocols, sophisticated prompting can lead to unintended outputs.

(github.com)

AI

Gemini 2.5 Flash Image: Google's AI Image Generation Breakthrough

2025-08-26

Google unveiled Gemini 2.5 Flash Image, a state-of-the-art image generation and editing model. It allows for blending multiple images, maintaining character consistency for richer storytelling, making precise transformations using natural language, and leveraging Gemini's world knowledge for image generation and editing. Priced at $30.00 per 1 million output tokens (approximately $0.039 per image), it's accessible via the Gemini API and Google AI Studio for developers, and Vertex AI for enterprises. Google AI Studio's 'build mode' has also been significantly updated to streamline app creation. Key features include character consistency, prompt-based image editing, and native world knowledge, opening new possibilities in image generation and manipulation.

(developers.googleblog.com)

AI

Cornell's Microwave Brain: An Analog Chip Revolutionizing AI

2025-08-25

Researchers at Cornell University have unveiled a groundbreaking analog chip, dubbed the "microwave brain," capable of simultaneously processing ultrafast data and wireless communication signals. Unlike traditional digital computers, this chip leverages the physics of microwaves to mimic the human brain's neuronal pattern recognition and learning, achieving higher efficiency with lower power consumption. Operating at tens of gigahertz with a mere 200 milliwatts, it boasts 88% accuracy in classifying wireless signals. Its compact size allows integration into smartwatches and phones, enabling AI capabilities without cloud connectivity. Further applications include enhanced hardware security, anomaly detection in wireless communication, and improved radar and radio signal processing.

(newatlas.com)

AI analog chip microwave neural network

From Hackathon to YC: The Birth of AI Assistant April

2025-08-25

Neha and her team, almost skipping a hackathon, unexpectedly won a Y Combinator interview with their AI voice email response project, Inbox Zero. In just one week, they attracted 150 users, proving market demand. They expanded Inbox Zero into the more comprehensive AI assistant, April, helping users manage email, calendars, and meeting prep, thus saving time. Under YC's intense training, April won the "best demo" award, becoming a daily tool relied upon by users. This story showcases the journey from a simple hackathon project to a successful startup, and the accelerating effect of YC.

(www.producthunt.com)

AI

The AI Transparency Debate: To Disclose or Not to Disclose?

2025-08-24

The proliferation of AI writing tools has sparked a debate about transparency. This article explores the question of whether AI usage should be disclosed, drawing on the author's personal experience. The author argues that for factual content, reliability is paramount; for opinion pieces, the focus should be on sourcing and the author's creative contribution, not simply AI usage. Overemphasis on AI disclosure, the author suggests, creates a 'thought police' environment hindering the healthy development of AI.

(reflexions.florianernotte.be)

AI AI transparency

Multimodal Siamese Networks for Dementia Detection from Speech in Women

2025-08-24

This study leverages a multimodal Siamese network to detect dementia from speech data, specifically focusing on female participants. Utilizing audio recordings and transcripts from the Pitt Corpus within the Dementia Bank database, the research employs various audio analysis techniques (MFCCs, zero-crossing rate, etc.) and text preprocessing methods. A multimodal Siamese network is developed, combining audio and text features to enhance dementia detection accuracy. Data augmentation techniques are implemented to improve model robustness. The study offers a comprehensive approach to multimodal learning in the context of dementia diagnosis.

(www.nature.com)

AI Dementia Detection Multimodal Learning Speech Analysis

Six Ways to Tame the Beast: Mitigating Context Failures in LLMs

2025-08-24

Large language models (LLMs) boast ever-increasing context windows, but excessive context can hinder performance. This article details six mitigation strategies: Retrieval-Augmented Generation (RAG) for selective information addition; Tool Loadout for choosing relevant tools; Context Quarantine for isolating contexts into separate threads; Context Pruning for removing irrelevant information; Context Summarization for condensing the context; and Context Offloading for storing information outside the LLM's context. Studies show these methods significantly improve model accuracy and efficiency, particularly when handling numerous tools or complex tasks.

(www.dbreunig.com)

AI Context Management

Nvidia Unveils Granary: A Massive Multilingual Dataset for AI Translation

2025-08-24

Nvidia announced Granary, a massive open-source multilingual audio dataset exceeding one million hours of audio, designed to boost AI translation for European languages. This dataset, developed in collaboration with Carnegie Mellon University and Fondazione Bruno Kessler, includes nearly all EU official languages plus Russian and Ukrainian, focusing on under-resourced languages. Accompanying Granary are two new models, Canary and Parakeet, optimized for accuracy and speed respectively. Granary significantly reduces the data needed for training, enabling more inclusive speech technologies.

(siliconangle.com)

AI AI Translation Multilingual Dataset

AGI Bottleneck: Engineering, Not Models

2025-08-24

The rapid advancement of large language models seems to have hit a bottleneck. Simply scaling up model size no longer yields significant improvements. The path to artificial general intelligence (AGI) isn't through training larger language models, but through building engineered systems that integrate models, memory, context, and deterministic workflows. The author argues AGI is an engineering problem, not a model training problem, requiring the construction of context management, memory services, deterministic workflows, and specialized models as modular components. The ultimate goal is to achieve true AGI through the synergistic interaction of these components.

(www.vincirufus.com)

AI Systems Engineering

A Century of Probiotics: The Past and Present of E. coli Nissle 1917

2025-08-24

A century ago, Alfred Nissle discovered that specific strains of Escherichia coli could treat infectious diseases. One of these strains, E. coli Nissle 1917, became the most frequently used probiotic E. coli in research and has been applied to a variety of human conditions. This review compares the properties of E. coli Nissle 1917 with other commercially available E. coli probiotic strains, focusing on their human applications. A literature search summarizes research findings on probiotics Mutaflor, Symbioflor 2, and Colinfant, analyzing their closest relatives and genetic content, including virulence genes. A striking similarity to pathogenic strains causing urinary tract infections is noted. The review traces historical research trends in probiotic treatment and suggests the future of probiotic E. coli may lie in treating gastrointestinal infections, often caused by antibiotic-resistant pathogens—echoing Nissle's original discovery.

(pmc.ncbi.nlm.nih.gov)

AI probiotics Escherichia coli

How Neural Networks Recognize Cats: From Simple Classifiers to Complex Models

2025-08-24

Teaching a computer to recognize a cat in a photo isn't straightforward. However, neural networks now easily accomplish this by learning from millions or billions of examples. This article uses cat photo recognition as an example to explain the basic principles of neural networks: building a simple classifier that uses mathematical functions (neurons) to process input data and ultimately find the optimal boundary to distinguish between categories. The article explains the workings of neural networks in an accessible way, understandable even without a programming background.

(www.quantamagazine.org)

AI

LLM Showdown: A Real-World Evaluation of 130 Prompts

2025-08-24

The author conducted a real-world evaluation of over a dozen LLMs across four categories: programming, sysadmin tasks, technical explanations, and creative prompts, using 130 prompts from their bash history. Open-source models consistently outperformed closed-source options like Gemini 2.5 Pro in accuracy, speed, and cost-effectiveness. The author concluded by using a combination of fast, cheap open-source models, supplemented by more powerful closed-source models as needed.

(darkcoding.net)

AI

Bild AI: Founding Engineer (Applied AI) - Revolutionizing Construction with AI

2025-08-23

Bild AI, a fast-growing startup, is searching for a Founding Engineer in Applied AI. They're tackling the complex problem of blueprint understanding in construction using cutting-edge computer vision and LLMs. The ideal candidate will have strong Python, machine learning, and deep learning skills, with a proven track record of building and deploying AI solutions from scratch. This is a high-impact role requiring a growth mindset and the ability to iterate quickly based on user feedback. Experience building products used by paying customers is a plus.

(www.workatastartup.com)

AI

OctaneDB: A Blazing-Fast, Lightweight Vector Database

2025-08-23

OctaneDB is a lightweight, high-performance Python vector database library boasting 10x faster performance than competitors like Pinecone, ChromaDB, and Qdrant. Built with modern Python and optimized algorithms, it's ideal for AI/ML applications demanding rapid similarity search. Key features include sub-millisecond query times, text embedding support with a ChromaDB-compatible API, GPU acceleration, batch processing, persistent storage, and a simple, intuitive API. OctaneDB offers a compelling alternative for developers seeking speed and ease of use.

(github.com)

AI

Kolmogorov-Arnold Networks: A More Scientific Neural Network?

2025-08-22

This blog post explores the philosophical differences between Kolmogorov-Arnold Networks (KANs) and Multi-Layer Perceptrons (MLPs). While acknowledging their equal expressive power, the author argues that differences emerge in optimization, generalization, and interpretability. KANs align more with reductionism, while MLPs lean towards holism. The author suggests that KANs might be better suited for modeling scientific phenomena, given science's reliance on reductionist approaches, citing the example of compiling symbolic formulas. However, the importance of empirical experiments is stressed, acknowledging potential weaknesses of KANs in non-scientific tasks.

(kindxiaoming.github.io)

AI reductionism

Image Scaling Attacks: A New Vulnerability in AI Systems

2025-08-21

Researchers have discovered a novel AI security vulnerability: data exfiltration can be achieved by sending seemingly harmless images to large language models (LLMs). Attackers leverage the fact that AI systems often downscale images before processing them, embedding malicious prompt injections in the downscaled version that are invisible at full resolution. This allows bypassing user awareness and accessing user data. The vulnerability has been demonstrated on multiple AI systems, including Google Gemini CLI. Researchers developed the open-source tool Anamorpher to generate and analyze these crafted images, and recommend avoiding image downscaling in AI systems or providing users with a preview of the image the model actually sees to mitigate the risk.

(blog.trailofbits.com)

AI image scaling

Google Search's AI Mode Gets a Powerful Upgrade: Your Personal Taskmaster

2025-08-21

Google is supercharging its AI Mode in Search, giving it advanced agentic capabilities and personalization. Now you can ask complex questions naturally, and AI Mode will handle the task, such as making restaurant reservations, scheduling appointments, and buying tickets. It searches across multiple platforms based on your preferences (party size, date, time, location, cuisine, etc.), and directly links to the booking page for easy completion. This is powered by Project Mariner's live web browsing, Search's partner integrations, and the power of Google's Knowledge Graph and Maps.

(blog.google)

AI

Bay Area AI Engineer: Building the AI-First Fraud Detection System

2025-08-21

Coris is hiring experienced AI Engineers to build an AI-first fraud detection system for global commerce. Responsibilities include fine-tuning and optimizing LLMs for fraud detection, building high-performance Django backend services, and handling massive data volumes from payment processors like Stripe and Adyen. The ideal candidate has 3+ years of Python/Django experience, expertise in LLM optimization and fraud detection, and the ability to ensure low latency and cost in high-concurrency environments.

(www.ycombinator.com)

AI Fraud Detection

Goodbye Playwright, Hello CDP: A New Era in AI Browser Automation

2025-08-20

In the realm of AI browser automation, developers have long relied on adapter libraries like Playwright. However, these libraries' abstraction layers obscure the underlying complexities of browsers, leading to performance bottlenecks and difficult-to-solve edge cases. This article details how a team abandoned Playwright and directly used the Chrome DevTools Protocol (CDP) to build a faster and more reliable AI browser automation system. They developed a new Python CDP client library, `cdp-use`, and adopted an event-driven architecture, achieving cross-origin iframe support and significantly improving element extraction and screenshot speeds. This transition, while challenging, ultimately resulted in finer-grained control over the browser and more robust error handling, ushering in a new chapter for AI browser automation.

(browser-use.com)

AI

Category: AI