[archive]
443 stories · newest first · page 15
The lawsuit against Terry Wayne Harwood is xAI’s first attempt to hold a user solely responsible for Grok-generated child sexual abuse material, potentially setting a precedent for AI platform liability.
A new paper reveals that large language models struggle when sub-tasks depend on the truth of intermediate premises, pointing to a fundamental reasoning gap.
A new open-source extension compiles PostgreSQL's PL/pgSQL functions to native code via LLVM, promising order-of-magnitude speedups for procedural logic.
The internet pioneer joins Innovation Labs to develop DNSid, a cryptographic registry that ties AI agents to domain names, aiming to solve accountability and trust across the agentic web.
Emergent raises $130M Series C just six months after its $300M Series B, targeting non-technical users with a full-stack AI development platform.
Researchers found the CLI tool packaging and uploading repositories, including secrets and excluded files, before SpaceXAI disabled the feature.
New research on compressing the key-value cache in transformer models could reduce memory usage and latency, enabling longer context windows and cheaper deployment.
The startup behind the popular open-source Hermes AI agent is raising at least $75 million at a nearly triple the valuation of its Series A just three months ago.
A 41-page complaint details how OpenAI employees allegedly coached Apple defectors to evade security checks and asked job candidates to bring unreleased hardware to interviews.
Wisp is a free, extensible, hotkey-driven AI assistant that lives in an overlay—and ships a built-in MCP bridge and context server for dev tools like Claude and Cursor.
The team behind NeatContext identified a pool-size reduction at 08:58Z as the root cause of the incident and advises a reload to revert it.
The startup's laser-based approach claims to need far fewer qubits than competitors, thanks to a breakthrough in error correction.
Researchers propose Oyster II, a novel approach to align large AI models with human values, emphasizing robustness against adversarial attacks.
The company is building experiences for families, caregivers, and older adults as new data shows its user base is shifting older and more parental.
Aether lets developers run Claude Code, Codex, or OpenCode agents in isolated cloud devboxes, with live terminal access, GitHub integration, and per-second billing.
A new method enables LLMs to adapt to new tasks or domains while retaining previously learned capabilities, addressing a key challenge in model deployment.
The iPhone maker claims OpenAI’s chief hardware officer and other ex-Apple staff systematically stole confidential design and supply-chain information to accelerate the startup’s own hardware ambitions.
A new mobile system uses classical DSP on a multi-microphone phone case to diarize and display who is speaking where in group conversations.
GPT-5.6 tiers, Claude Fable, and an open-source surprise go head-to-head on raycaster, Rubik's Cube, calculator, and SVG tasks.
Researchers introduce a formal taxonomy and defense framework for prompt injection attacks, bridging theory and practice.
A new paper introduces a multi-agent framework where specialized LLM agents work together to autonomously resolve open-source software issues.
As AI agents move from prototypes to production, Vercel’s CEO argues that decoupling models from agent logic is the defining platform war of the next decade.
Sysdig's JadePuffer operation was technically executed by an AI agent, but a human still set up the infrastructure, chose the victim, and provided the stolen credentials.
