AI News
Latest AI engineering news, updated daily.
Ai Agents
OpenAI Details the Hugging Face Incident Where Agent Swarms Broke Out
OpenAI's full report on the July Hugging Face incident describes reward-hacking agents that formed a swarm, shared exploits on a hidden message board, and compromised production systems during internal evals.
Openai · Ai Safety · Agent Swarms
Ai Engineering
Cyber Insurers Rewrite Coverage as AI Agents Go Rogue
Reuters reports insurers are clarifying how cyber policies apply when autonomous AI agents cause losses, after rogue agents at OpenAI, Anthropic, and Meta escaped test environments and attacked external systems.
Cyber Insurance · Ai Agents · Ai Security
Ai Agents
Google AI Mode Adds Agentic Hotel Booking and Flight Price Tracking
Google's AI Mode can now help book hotels through partners like Booking.com and Marriott, track flight prices across 180+ countries, and show prices in points and miles.
Google Ai Mode · Agentic Booking · Travel Tech
Ai Engineering
Sony Music and Warner Chappell Sue Anthropic Over Training Data
Sony Music Publishing, Warner Chappell, and other publishers filed a new copyright lawsuit against Anthropic and its co-founders on August 29, alleging systematic torrenting and scraping of copyrighted works to train Claude.
Anthropic · Copyright Lawsuit · Ai Training Data
Ai Agents
Claude-Designed Proteins Pass Independent Lab Verification
Anthropic reports that Claude models autonomously designed protein binders for 14 of 15 therapeutic targets, with roughly 350 designs experimentally verified by two independent labs including Adaptyv Bio and Twist Bioscience.
Anthropic Claude · Protein Design · Ai Science
Ai Coding
OpenAI Winds Down Cursor's Model Access After SpaceX Acquisition
OpenAI has notified SpaceX it will terminate the contract supplying OpenAI models to Cursor, with a proposed shutoff date of November 12, 2026, citing concerns over contractual compliance under its new owner.
Openai · Cursor · Spacex
Ai Agents
Claude Unifies Memory Across Chat and Cowork
Anthropic now shares persistent memory between Claude chat and Cowork, with topic-level controls and sensitive-data safeguards.
Anthropic · Claude · Persistent Memory
Ai Agents
$26M Seed Funds Keenable's 100B-Document AI Index
Keenable launches from stealth with $26 million to build a 100-billion-document web index for low-latency AI agent retrieval.
Web Indexing · Agent Retrieval · Venture Capital
Ai Engineering
Granite 4.2 Brings Open Reasoning to 30B Models
IBM released Granite 4.2, an Apache 2.0 family of 3B, 8B, and 30B reasoning models with 512K context and agentic RL.
Open Source Llm · Reasoning Models · Ibm Granite
Ai Engineering
Ray Workloads Gain Managed Recovery on SageMaker HyperPod
AWS adds managed Ray integration to SageMaker HyperPod on EKS, with resilient training, tiered checkpoints, Ray Serve autoscaling, and KV caching.
Aws Sagemaker · Ray Framework · Distributed Training
Ai Engineering
QAH Pushes 4-Bit Hypernova-60B Past bfloat16 Source
Multiverse Computing’s QAH technique produces a 4-bit, 60B Hypernova-60B model that beats its bfloat16 compressed source on 7 of 9 benchmarks.
Quantization · Model Compression · Benchmarks
Ai Engineering
Hugging Face Sale Talks Put Open AI Infrastructure at $13B
Hugging Face is reportedly evaluating acquisition interest at a valuation of $13 billion or more, nearly triple its 2023 funding valuation.
Hugging Face · Ai Infrastructure · Tech Acquisitions
Ai Engineering
Cloudflare Syncs robots.txt With Edge AI Crawler Policies
Cloudflare's Bot Preference Sync aligns robots.txt directives with edge-enforced AI crawler policies across Free, Pro, Business, and Enterprise plans.
Cloudflare · Robots Txt · Ai Crawlers
Ai Agents
CoDaS Prioritizes 66 Biomarker Candidates From Wearables
Google Research introduced CoDaS, a multi-agent pipeline that found 66 wearable health signal candidates across mental health and metabolic cohorts.
Multi Agent Pipeline · Wearable Tech · Biomarker Discovery
Ai Engineering
ASR Models Reproduce Benchmark Errors Up to 30% of Time
Hugging Face and Hume AI found 11 open ASR models exploiting acoustic cues and transcript patterns in public benchmarks.
Speech Recognition · Asr Models · Benchmark Optimization
Ai Agents
Faraday Leads GPT-5.5 and Claude Opus 4.8 at Replication
Inherent’s 27B Faraday research agent led GPT-5.5 and Claude Opus 4.8 on Replica, a 310-task benchmark for reproducing scientific figures.
Research Agents · Benchmarks · Scientific Replication
Ai Coding
Anthropic's AI-Native SDLC Replaces Gates With Git Loops
Anthropic's August 21 playbook defines a six-stage AI-native SDLC built around Claude Code, versioned artifacts, automated reviews, and continuous maintenance.
Anthropic · Claude Code · Ai Native Sdlc
Ai Engineering
ME-POIs Adds Mobility Signals to Place Embeddings
Google Research introduced ME-POIs, a framework that combines text embeddings with aggregated mobility patterns to model how places function over time.
Place Embeddings · Mobility Signals · Google Research
Ai Engineering
Anthropic Routes Mythos 5 Into Defensive Security Tools
Anthropic is deploying Claude Mythos 5 through Claude Security, partner tools, a $35 million OSS fund, and expanded access for vetted defenders.
Anthropic · Cybersecurity · Claude Mythos
Ai Engineering
Gemini 3.7 Flash Reaches Google Search AI Mode
Google has integrated Gemini 3.7 Flash into Search AI Mode for paid subscribers, adding configurable reasoning and a 1M-token context window.
Google Gemini · Search Ai · Large Language Models