Thoth: Your New Local-First AI Assistant
Thoth is an open-source AI assistant that runs on your own devices, keeping your data private. It's designed to be a simple, powerful alternative to cloud-based AI helpers.
Archive
All 3,502 AI stories, newest first · page 118 of 146
Testing AI Agents That Never Give the Same Answer Twice
Testing AI agents is tricky because they often produce different answers. Researchers are developing new methods to evaluate their performance consistently. This matters for everyone who relies on AI tools for daily tasks.
Oscars Ban AI-Generated Performances and Scripts
The Academy of Motion Picture Arts and Sciences has ruled that AI-generated actors and scripts are ineligible for Oscars. This decision aims to preserve the human element in filmmaking.
via TechCrunch AI#Filmmaking
New Open-Source Tool Lets You Test AI Agents Like a Pro
TrainForgeTester is a new open-source tool that helps you test AI agents in real-world scenarios. It focuses on catching mistakes like wrong tool calls or skipped steps, making AI agents more reliable for everyday use.
Mnemory: Giving AI Agents a Memory to Remember
Mnemory is a new tool that lets AI agents remember information across conversations. This could make AI assistants much more useful in everyday tasks.
How OpenAI Keeps ChatGPT Safe for Everyday Users
OpenAI outlines its efforts to ensure ChatGPT remains safe through model safeguards and expert collaboration. These measures protect users from harmful content and misuse.
How Kepler Built Verifiable AI for Financial Services
Kepler used Claude to create AI systems that can explain their decisions, which is crucial for financial services. This helps build trust in AI-driven financial tools. In plain English, it means AI can now show its work, making it more reliable for everyday banking and investments.
How AI Helped Migrate 6,000 React Tests in Days
Developers used AI tools to quickly update thousands of outdated React tests. This approach could make software maintenance faster and less painful for teams.
Duralang: Simplifying AI Agent Development with One Decorator
Duralang is a new tool that makes building AI agents easier by turning every LangChain call into a Temporal Activity. This could make AI agents more reliable and easier to manage for developers.
DeepSeek V4 Pro Matches GPT-5 in NIST's Rigorous AI Test
A new evaluation by NIST shows that DeepSeek V4 Pro performs as well as GPT-5 in key AI benchmarks. This could mean more competition and better options for AI users.
Codex: OpenAI's AI Tool for Automating Work Beyond Chat
Codex is a powerful AI model that automates tasks, connects tools, and creates real outputs like documents and dashboards. It's designed to go beyond simple chat, helping users get tangible results.
ASU Uses AI to Create Courses from Professors' Work Without Their Knowledge
Arizona State University is using AI to generate courses from professors' materials without their consent. This raises concerns about academic integrity and ownership of intellectual property.
AI that writes and peer-reviews its own research papers
A new AI called Cajal can draft research papers and simulate peer review. This could speed up scientific discovery but raises questions about authenticity.
AI Outperforms Doctors in Emergency Room Diagnoses, Harvard Study Finds
A Harvard study found that certain AI models provided more accurate diagnoses in emergency room cases than human doctors. This suggests AI could become a valuable tool in fast-paced medical settings.
AI Learns to Generate Secure Random Numbers Like a Computer
Researchers have trained an AI model to generate pseudorandom numbers, mimicking how computers create seemingly random sequences. This could lead to more secure systems and new ways to test randomness.
AI-Generated Music Overwhelms Streaming Services — But Is Anyone Listening?
Streaming platforms are flooded with AI-generated music, raising questions about its quality and demand. The surge highlights a growing divide between what AI can create and what listeners actually want.
TabPFN Shows Promise in Predicting Alzheimer's Progression from Limited Data
Researchers evaluated TabPFN against traditional machine learning methods for predicting Mild Cognitive Impairment (MCI) to Alzheimer's Disease (AD) conversion. The study found TabPFN performed comparably to traditional models despite limited longitudinal data.
Stripe's Link Now Lets AI Agents Handle Your Payments
Stripe has updated its Link digital wallet to allow AI agents to make purchases on your behalf with your approval. This could make online shopping and subscriptions easier and more secure.
Step-Level Optimization Cuts Costs for Computer-Use AI Agents
Researchers propose a new approach to optimize compute resources for GUI-interacting AI agents, reducing costs and improving efficiency. The method targets long-horizon tasks where uniform compute allocation is inefficient.
SimplePDF Copilot: AI-Powered PDF Form Filling Without Cloud Uploads
SimplePDF Copilot is a new AI tool that lets users fill PDF forms directly in the browser without uploading documents to the cloud. The tool integrates with any LLM and has already gained significant traction.
Pentagon partners with Nvidia, Microsoft, and AWS for classified AI deployment
The Department of Defense has signed agreements with Nvidia, Microsoft, and AWS to deploy AI on classified networks. This move follows a dispute with Anthropic over AI model usage terms, highlighting the DOD's push for vendor diversification.
OpenAI's Privacy Filter Helps Developers Build Safer Web Apps
OpenAI has released a new tool to help developers build web apps that protect user privacy. This filter ensures sensitive data isn't exposed in AI interactions.
OpenAI Restricts Access to Cybersecurity Tool GPT-5.5 Cyber
OpenAI is limiting access to its new cybersecurity tool, GPT-5.5 Cyber, to only critical cyber defenders. This follows criticism of Anthropic for restricting its own AI tool, Mythos.
OpenAI boosts ChatGPT security with new Yubico partnership
OpenAI is introducing optional advanced security features for ChatGPT accounts, including a collaboration with Yubico for physical security keys. This move aims to make user accounts more secure against unauthorized access.