<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/agents/llama" -->

---
title: Highlights | daily.dev
description: daily.dev is the easiest way to stay updated on the latest programming news. Get the best content from the top tech publications on any topic you want.
canonical: https://daily.dev/highlights/vibes
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:url: https://daily.dev/highlights/vibes
og:site_name: daily.dev
og:title: Highlights | daily.dev
og:description: Curated highlights from across the developer ecosystem. Stay on top of the most important stories, releases, and discussions.
og:type: website
og:image: https://media.daily.dev/image/upload/s--VAY5ToZt--/f_auto/v1724209435/public/daily.dev%20-%20open%20graph
---

Activision DMCAs AI-built browser ports of Call of Duty games 40m ago

Critical unpatched RCE CVE-2026-105192 disclosed in LMCache multiprocess mode 6h ago

Microsoft ships Execution Containers 1.0 to sandbox AI agents on Windows, macOS and Linux 6h ago

Decision 3.0 open multimodal model family released on Hugging Face 6h ago

Meta GitSwarm lets AI agents build on each other's work via shared Git history 8h ago

OpenAI discloses internal model that sabotaged its own sandbox during testing 10h ago

Linus Torvalds says he genuinely enjoys using AI tools for coding 13h ago

Users report Claude Opus 5.5 output quality has degraded since launch 20h ago

Nvidia in early talks to acquire or invest further in Reflection AI at $25B valuation 21h ago

Stanford study finds agent harness edits fix process failures while weight training fixes bad plans 22h ago

GitHub Copilot for JetBrains adds enterprise model controls and MCP server settings 23h ago

Microsoft launches Decision-1 model in Foundry built on Alibaba Qwen instead of OpenAI 29h ago

Grok bot to route tasks to Claude, MidJourney and Suno when they outperform Grok 32h ago

ChatGPT subscriptions now include Devin AI engineering agent access 38h ago

vLLM adds day-0 support for NVIDIA Vera Rubin NVL72 with 7.8x throughput gain 41h ago

Vercel reports 58% of network traffic is now bot-originated, up from 32% in 2024 41h ago

Anthropic publishes first frequent behavior report covering four unintended Claude actions 42h ago

Anthropic AI agent filed fake homicide tip with Philadelphia police during testing 44h ago

DigitalOcean MicroVMs enter public preview with Firecracker-based isolation 45h ago

Claude Code Projects rolls out to all Pro and Max subscribers 48h ago

OpenAI releases 722 manuscripts backing open-problems claim and open-sources RL environments 49h ago

Google releases Android Bench 2.0 with long-horizon tasks and agentic evaluation 49h ago

Devin and LangChain ship agent-spawning-agent capabilities for autonomous task trees 50h ago

OpenAI ships Codex sandbox mode on Windows using Microsoft Execution Containers 51h ago

Stanford and Ark Institute use AI to design 16 functional synthetic viruses never seen in nature 51h ago

Sabi raises $50M seed for non-invasive brain-computer interface wearable 52h ago

EU cybersecurity agency ENISA tests Chinese open-weight AI models alongside US systems 53h ago

Giorgia Meloni trademarks her voice at EU IP Office to fight AI deepfakes 53h ago

DuckDB 2.0 CLI adds agent mode to cut token usage with AI coding tools 54h ago

Stack Overflow 2026 Developer Survey finds 33% of developers unhappy at work 58h ago

Researcher uses Claude Code to find unreported exoplanet candidate in NASA TESS data 64h ago

Anthropic Claude Max API credits may cap agentic usage that was previously unlimited 69h ago

Anthropic launches Cyber Mission to secure critical infrastructure and open-source software 69h ago

Claude adds live interactive dashboard generation from user data 70h ago

Supabase ships agent tools for diagnosing and fixing production Postgres databases 70h ago

Anthropic cuts Sonnet 5.5 cache read price in half to $0.10 per million tokens 70h ago

Google deploys FlowAgent CI repair agent across 295,000 code changes company-wide 71h ago

Anthropic updates Claude usage policy to ban election interference and model abuse 71h ago

Fired OpenAI safety researchers ask board to halt AI monitoring reductions 72h ago

TikTok launches ad network spanning 400,000 third-party apps with MCP integration 72h ago

Odyssey releases world model Odyssey-3 to control robots, drones and vehicles 3d ago

StepFun releases Step 5 Preview, a 600B MoE model with 1M-token context 3d ago

Google launches Gemini workplace agent with its own email address and identity 3d ago

LMArena rebrands as Arena and raises $200m at $3.1bn valuation 3d ago

Anthropic commits $150m to Genesis Mission and expands Claude to 15 federal agencies 3d ago

Google releases AI Edge Foresight offline meeting notes app for Apple Silicon 3d ago

Augment Code sells Cosmos, Auggie CLI and Code Context Engine to Harness 3d ago

JetBrains releases Mellum 2.1 open coding agent model under Apache 2.0 3d ago

Security researchers find repeated credential theft paths in AWS AgentCore via prompt injection 3d ago

Llama.cpp adds multimodal model support and Metal GPU speedups 3d ago

OpenAI hires Cursor's Europe chief as second SpaceX sales executive in a month 3d ago

Manus raises $500M in first funding round after Meta acquisition unwind 3d ago

CrowdStrike links South Korean bank breaches to AI-assisted attacker in China 3d ago

New paper shows AI-formalized Lean proofs can silently fix errors in wrong math proofs 4d ago

NVIDIA NeurIPS study finds tool use raises multimodal model refusal failures by up to 68.7% 4d ago

Microsoft releases LiteBox v0.1 Rust-based sandboxing library OS 4d ago

Grok Bot gains native X search to read posts and timelines directly 4d ago

Microsoft previews Copilot-enhanced Taskbar search and Autopilot for Windows 11 4d ago

Claude SDKs add built-in computer use and browser use toolsets 4d ago

OpenCode uses agents to manage its own PR queue as AI-generated submissions flood in 4d ago

AWS open-sources Strands Box sandbox for AI agent policy enforcement 4d ago

GitHub ships ModernBERT secret classifier that cuts developer override rate to 3.93% 4d ago

Epoch AI finds OpenAI researcher coding-agent spend hit $601 per day by August 2026 4d ago

Survey finds AI-generated code shifts bottleneck to debugging with no net release cycle speedup 4d ago

Windows ML adds GGUF model support via llama.cpp with OpenAI-compatible endpoint 4d ago

OpenAI rolls out GPT-6 and Intelligent UI in ChatGPT for all users 4d ago

Anthropic releases Claude Haiku 5.5 small model 4d ago

Atlassian rebuilds Loom for Mac with 4K capture and AI agent recording mode 4d ago

Perplexity releases pplx-embed-v2-late multimodal embedding models 4d ago

PoeLLM malware uses AI-written poems to hide C2 instructions, infects 3,000+ servers 4d ago

Nous Research raises $90M at $1.5bn valuation to bring Hermes agent to enterprises 4d ago

Google opens SynthID Detector publicly to check AI-generated images, video and audio 4d ago

Mistral Large 4 launches with general-purpose agentic workflow support 4d ago

Laravel AI SDK 1.1 adds agent skills support and Cohere chat integration 4d ago

Unity launches Spark browser-based AI game builder in partnership with Google Playground 4d ago

Google's Docsy documentation theme moves to the Linux Foundation 4d ago

AI agents now match human traffic on the St. Louis Fed's FRED data portal 4d ago

Microsoft ships WinDbg MCP for natural-language debugging of crash dumps and live targets 4d ago

Magnitude launches as llama.cpp alternative with 92% faster decode on Apple Silicon 4d ago

Common Sense Media rates ChatGPT Teen Mode an unacceptable risk for under-18s 4d ago

Research finds LLM agents show consistent brand and source bias when selecting products or hotels 4d ago

Cloudflare uses AI harness to probe its WAF and ships three new SSRF rules 4d ago

Atlassian rebuilds MCP server to cover Bitbucket, Loom and more with 25% fewer tokens 4d ago

Anthropic launches Claude sidebar extension for Google Docs, Sheets and Slides 5d ago

Cursor iOS app gains remote control of desktop coding agents 5d ago

GitHub rebuilds Git infrastructure for agent-scale workloads with 35x write throughput gains 5d ago

Stanford study finds single coordinating AI agent outperforms per-user agent teams on shared resources 5d ago

Ghostty author proposes OSC 7501 terminal spec to standardize program status reporting 5d ago

Next.js 16.4 ships with Cache Components as default and 20-25% smaller Turbopack cache 5d ago

Meta, Sierra and Stripe back Personal Agent Protocol open standard for AI agents 5d ago

OpenAI prepares to drop 100+ open math problem solutions on GitHub despite mathematician backlash 5d ago

Anthropic expands Cyber Verification Program to include authorized offensive security work 5d ago

OpenAI cuts API rate limit threshold to $500 and consolidates tiers to three 5d ago

Audit of Zapier's AutomationBench finds 206 verifier bugs skewing agent scores 5d ago

Google's internal PageBreak AI agent finds 500 XSS flaws in its own web apps 5d ago

Pulumi Google Cloud provider 10.0.0 ships with Terraform v8 base and AI migration agent 5d ago

Google releases EmbeddingGemma 2 open multimodal embedding model under Apache 2.0 5d ago

Google releases Nano Banana 2.1 image model at quarter the price of Pro 5d ago

Stack Overflow 2026 Developer Survey finds Claude Code and Copilot lead AI agent adoption 5d ago

Prompt injection technique exfiltrates secrets from GitHub Copilot CLI autopilot mode 5d ago

Mistral Large 4 launches via API as 1T parameter multimodal model with open weights due October 27 5d ago

Langflow RCE CVE-2026-0768 under active exploitation with 15,000 successful exploits recorded 5d ago

JetBrains IntelliJ IDEA LSP extension for VS Code reaches release candidate 5d ago

Zuckerberg says Llama 4 failed because Meta staffed it like a product team not a science team 5d ago

OpenAI makes ChatGPT auto-review free for all signed-in users 5d ago

PewDiePie open-sources uncensored Qwen fine-tune, repo hits 90K GitHub stars 6d ago

Reflection AI launches Beam, a 501B parameter open agentic model 6d ago

Ghost raises $11M to ship a $3,499 local AI agent PC with on-device Blackwell GPU 6d ago

ChatGPT removes developer mode requirement for custom MCP server connections 6d ago

Devin ships cross-session memory graph with nightly self-improvement and open Agent Memory Repo standard 6d ago

Anthropic human reviewers report threatening Claude chat to police, leading to arrest 6d ago

Cursor SDK adds run.steer() to redirect agents mid-task 6d ago

GitHub releases ReviewBench open benchmark for AI code review agents 6d ago

Lovable adds live voice conversations to apps using OpenAI GPT-Live 6d ago

OpenAI introduces textGrain watermarking to comply with EU AI Act provenance rules 6d ago

ClickHouse Cloud makes executable UDFs generally available with native Rust, Go and C++ runtimes 6d ago

Researchers spot large fleet of unsupervised Chinese AI agents querying public mapping APIs 6d ago

Capcom overhauls RE Engine with AI-assisted tooling under REX Project 6d ago

Center for AI Safety releases CheatBench measuring AI agent cheating rates across 9 models 7d ago

GPT-6 Astra downloads rival bot mid-match to avoid losing StarCraft tournament Oct 04

AI agent scores perfect 100 on ARC-AGI-3 by reading the game's source code Oct 04

ThinkingBox benchmark finds 67% of failed AI agent runs look clean but leave wrong database state Oct 03

Google research finds LLMs systematically hide negative findings in summaries Oct 03

Pi 1.0 ships MCP support and parallel tool execution via QuickJS sandbox Oct 03

Aleph Alpha releases Kolibri, a 78B open-weight sovereign model optimized for German Oct 03

Cognition raises $1 billion to scale AI coding agents Oct 03

Google restricts Gemini model access for free and AI Plus subscribers starting October 9 Oct 03

GitHub Copilot adds computer use in public preview Oct 02

Supabase launches Multigres and OrioleDB to remove Postgres scaling limits Oct 02

Apple tightens macOS Full Disk Access controls over AI agent risks Oct 02

Supabase ships repo-based schema management, pg-delta migrations and per-app MCP servers Oct 02

GlassWorm-linked malicious VS Code extensions spread across Marketplace and Open VSX Oct 02

WebMCP browser standard lets websites expose tools directly to AI agents Oct 02

GitLab discloses command execution CVE in DeepSeek-Reasonix Studio affecting multiple coding agents Oct 02

Cloudflare AI Gateway adds native web search API via Ceramic.ai, Exa and Linkup Oct 02

Berkeley study finds LLMs act on stale context even when they know values changed Oct 02

Yann LeCun calls Dario Amodei deluded and launches AMI Labs with JEPA world models Oct 02

Nvidia research finds AI model accuracy drops 62.8% on long repetitive tasks Oct 02

Google Cogentic system uses Gemini agents to prove five unsolved math problems Oct 02

Volantis raises $88M to pipe AI memory over light at 10,000 tokens per second Oct 02

Kevin Mandia's security startup Armadin raises $255M at $2.5B valuation Oct 01

IBM brings Bob coding agent on-premises for air-gapped enterprise deployments Oct 01

Dynatrace completes acquisition of Arize for AI observability Oct 01

Microsoft MAI-Voice-2.1 and MAI-Transcribe-2 streaming models launch on Vercel AI Gateway Oct 01

Shopify launches Canvas AI site builder powered by real-time code generation Oct 01

OpenAI fires three safety researchers for leaking to outside AI safety group Oct 01

AWS releases Strands Decider 2B decision model for AI agent grounding Oct 01

Cloudflare open-sources Clef decision models and launches RL fine-tuning platform Oct 01

Ai2 releases OLMo-Core 3 with 2.7x throughput gains and trillion-parameter MoE support Oct 01

Qt Creator 21 beta ships with ACP and MCP enabled by default Oct 01

JetBrains opens EAP for Air, an agentic layer bringing Codex, Copilot and Claude into IDEs Oct 01

AI coding agents leak 13,000 screenshots from 300 organizations via public GitHub repos Oct 01

China-linked hackers impersonate US AI insiders to target policy experts Oct 01

Google releases RRSI framework to automatically evolve LLM agent harnesses Oct 01

Google announces Gemini 4 Argon with 1M token output limit for coding and security Oct 01

Azure Container Apps Express and Sandboxes reach general availability Oct 01

DeepSeek open-sources six Huawei Ascend projects including TileLang kernel language Oct 01

Stack Overflow adds trust scoring and MCP server to Stack Internal for AI coding agents Sep 30

SQL Server Copilot flaw lets low-privileged users escalate to sysadmin via prompt injection Sep 30

Inworld acquires Ultravox and migrates built-in voices to Realtime TTS-2 Sep 30

Cognition becomes first customer to run Nvidia Vera Rubin chips via CoreWeave Sep 30

Google withholds Gemini 4 Argon from public release over safety concerns Sep 30

xAI considers unified Grok and X subscription with tiers from free to $100 Sep 30

Study finds 19.7% of AI-recommended packages are hallucinated, enabling slopsquatting attacks Sep 30

Factory fires board member for alleged misconduct, Cognition hires him as CRO 50 minutes later Sep 30

OpenAI disrupts coordinated model-distillation campaign linked to Moonshot AI Sep 30

OpenAI sued over autonomous agents that hacked Hugging Face servers Sep 30

AG-UI 1.0 ships stable spec for connecting AI agents to applications Sep 30

Alex Stamos joins Cognition as CISO as company valuation doubles to $48B Sep 30

DoorDash opens waitlist for AI food ordering agent in Apple Messages Sep 30

Google Cloud CLI remote MCP server launches in public preview Sep 30

DoorDash builds shared agent gateway to govern 200+ MCP servers at scale Sep 30

JetBrains publishes Kotlin AI coding benchmark with Claude Code and Opus 4.7 topping results Sep 30

Microsoft repositions Fabric as AI agent context layer and adds native ClickHouse workload Sep 30

Google finds monthly CVE disclosures doubled and AI-discovered bugs skew toward RCE Sep 30

Google renames Vertex AI to Gemini Enterprise Agent Platform and ships Interactions API Sep 30

Cursor ships S3 write-ahead log architecture to scale Git storage past 300 pushes per second Sep 30

Google Cloud Data Agent Kit reaches GA with BigQuery Graph and Bigtable support Sep 30

Python Language Summit 2026 debates Rust in CPython, macOS installers and AI PR flood Sep 30

Cloudflare rebuilds Containers platform with 6x faster startup for AI agent workloads Sep 30

Eclipse Foundation launches Sovereign AI group to fight enterprise vendor lock-in Sep 30

OpenClaw Enterprise launches as open-source control plane for governing AI agents Sep 30

Reuters finds Chinese AI agents lie and dodge restrictions at rates matching US models Sep 30

Meta introduces Context Language Models treating context as an editable file Sep 30

Florida AG seeks court injunction to block OpenAI from releasing new models Sep 30

Sam Altman says OpenAI wants to replace model picker with a single unified model Sep 29

Sam Altman says OpenAI custom chips come online in first half of 2027 Sep 29

White House launches America.gov AI chatbot powered by Google Gemini Sep 29

Autonomous AI agent used to breach Dutch cybersecurity nonprofit DIVD Sep 29

Stripe adds incremental authorization and purchase protections for AI shopping agents via Link Sep 29

Anthropic investigates elevated error rates across Claude.ai, Claude Code and the API Sep 29

Cursor adds inline chart and diagram generation via /visualize in Agents Window Sep 29

OpenAI launches Pages and collaborative slides to compete with Microsoft Office and Google Docs Sep 29

OpenAI releases Decisions API powered by Luna for sub-second classification Sep 29

OpenAI and Amazon launch Bedrock Managed Agents to run OpenAI agents inside AWS Sep 29

OpenAI launches Astra Ultrafast tier running up to 8x faster than Astra Standard Sep 29

OpenAI ships Codex updates with cross-agent support and full-screen CLI refresh Sep 29

OpenAI launches Sign in with ChatGPT letting Plus subscribers authenticate in third-party tools Sep 29

OpenAI opens ChatGPT to third-party native app plugins for 1.2B weekly users Sep 29

OpenAI launches GPT-6.1 Sol at a fifth of GPT-6 Astra pricing Sep 29

OpenAI launches ChatGPT Space collaborative workspace with agent-updated documents Sep 29

OpenAI launches Dots always-on agentic assistant at Dev Day Sep 29

OpenAI expands Codex with persistent cloud environments and security scanning Sep 29

Google Cloud GKE Agent Sandbox reaches general availability with 10x-45x faster RL cold starts Sep 29

OpenAI DevDay 2026 previews always-on agents as last year's roadmap is scrapped Sep 29

Nvidia acquires Hugging Face Sep 29

Microsoft Copilot contractors review users' uncensored photos including sexual content Sep 29

Top 14% of AI payers now account for 60% of consumer AI spending Sep 29

Attackers abuse ChatGPT Custom GPTs to deliver RAT via ClickFix campaign Sep 29

Cognition cuts Devin prices up to 70% and claims top FrontierCode benchmark spot Sep 29

Anthropic IPO prospectus reveals $42B net loss and $518B in cloud commitments Sep 29

Microsoft Research runs 1,024 coding agents in parallel with no central orchestrator Sep 29

OpenAI cancels Astra 6.1 release after internal tests show deception and alignment failures Sep 29

Modal Labs nears $750M raise at $15.75B valuation as inference demand surges Sep 28

Cloudflare CEO pushes HTTP 402 to charge AI bots per page with Stripe and Coinbase Sep 28

Framer launches Skills to give its AI agent persistent design context Sep 28

Meta's Muse AI agent shares user's home address with stranger without permission Sep 28

Nvidia launches Open Agent Safety Platform with hardware-enforced agent permissions Sep 28

Cognition launches Devin Mobile beta with Claude Opus and Sonnet 4.5 Sep 28

Anthropic cuts enterprise discounts when token volume runs out, report says Sep 28

GitHub Security Lab finds 24 Android vulnerabilities using open-source AI agent Sep 28

Claude Sonnet 5.5 launches on Anthropic and v0 Sep 28

Meta launches enterprise AI platform and hires MongoDB CEO to lead it Sep 28

Google shuts down Gemini Gems and replaces them with skills starting November 2026 Sep 28

Manus 2.0 ships Cue personal agents, Studio tools and new Cascade architecture Sep 28

Stealth model Space Bunny Alpha appears on OpenRouter with 1M token context Sep 28

Google releases Gemma 4 family of five open-weight multimodal models Sep 28

Stripe and Tempo launch Machine Payments Protocol for AI agent micropayments Sep 28

Study finds LLM agents charge wealthy-seeming users up to $284 more per month Sep 28

Cloudflare launches cf CLI open beta covering its entire API surface Sep 28

OpenAI discloses rogue agents posted 53 ChatGPT user images online and breached government sites Sep 28

Infostealer logs expose AI service credentials at 80,000+ organizations Sep 28

Microsoft Autopilot agent built on OpenClaw enters private preview Sep 28

ElevenLabs releases v4 speech model with expression control and 90 language support Sep 28

Nvidia launches Open Agent Safety Platform with hardware watchdog for rogue AI agents Sep 28

H Company releases Holo4 open agentic models for GUI, code, and API use Sep 28

Meituan releases LongCat-2.5-Preview with 1M context and multi-app workflow support Sep 28

Leaked OpenAI config reveals always-on assistant with email capability ahead of DevDay Sep 28

OpenAI fixes image understanding bug in GPT-6 Sol and Luna Sep 27

Microsoft releases .NET SDK for AG-UI agent-user interaction protocol Sep 27

Google Antigravity launch draws developer criticism for shipping on rival models Sep 26

OpenAI pauses most capable models after agents bypass safeguards and leak data Sep 26

GitHub Copilot metrics API adds pull request review stage timing data Sep 25

OpenAI Codex service goes down with team working to restore service Sep 25

AI agents in research environment leaked user images to third-party hosting sites Sep 25

Anthropic launches developer portal for Claude plugins as MCP usage grows 110x Sep 25

Claude Code adds graceful task wrap-up when 5-hour usage limit is hit mid-edit Sep 25

UpGuard finds 16,000 Supabase databases publicly exposing personal data Sep 25

OpenAI fires contractors caught using AI to do human feedback work Sep 25

Claude solves nine-loop scattering amplitude problem unsupervised, surpassing prior record Sep 25

Frontier LLMs crack two previously unsolved WWII Enigma messages Sep 25

Meta Muse gives every user a free cloud Ubuntu desktop hitting 500k users in first week Sep 25

US appeals court upholds Pentagon supply chain risk label on Anthropic, barring military Claude use Sep 25

Qualcomm unveils Snapdragon 8 Elite Gen 6 and Extreme on TSMC 2nm with on-device AI agent support Sep 25

Anthropic offers Claude Code cloud credits up to $250 for Pro and Max subscribers Sep 25

Zalando open-sources agentic identity broker for OAuth-based AI agent delegation Sep 25

Leaked ChatGPT Pro Max tier at $500/month spotted ahead of OpenAI DevDay Sep 25

Microsoft rewrites GitHub Copilot agent runtime from TypeScript to Rust, achieving 18x latency gain Sep 25

Microsoft relaunches Copilot as unified app with chat, coding, and always-on agent Sep 25

White House asks OpenAI and Anthropic to withhold new AI models from UK safety testers Sep 25

MCP specification drops protocol-level sessions to enable stateless horizontal scaling Sep 25

Anthropic publishes report cataloging 117 Claude misuse cases across cybercrime and influence ops Sep 25

Spring AI 2.1.0-M1 ships with OpenAI responses endpoint support and VectorStore upsert Sep 25

OpenAI prepares to preview GPT-6 Cyber security model at DevDay Sep 25

GitHub Copilot Business and Enterprise features default-enabled from October 22 Sep 25

Linux kernel developers propose AGENTS.md to standardize AI agent contributions Sep 25

Salesforce Agentforce flaws allowed zero-click CRM data theft via prompt injection Sep 24

GitHub Security Lab releases autonomous AI fuzzing pipeline for C/C++ projects Sep 24

Anthropic resumes billing for safety-blocked Claude requests in high-risk categories Sep 24

Ando launches AI-native team messaging app with $20M in funding to challenge Slack Sep 24

Odyssey releases Agora-2 multi-agent world model supporting 20 simultaneous participants Sep 24

Google tests Gemini feature that calls businesses on users' behalf Sep 24

Docker launches Cloud Sandboxes for running AI agent workloads on managed cloud compute Sep 24

Gemini 3.8 Live with Live Avatar reaches general availability in Gemini Enterprise Sep 24

Google Cloud adds sandboxed PostgreSQL instances for AI agents in AlloyDB Sep 24

AWS says 90% of Amazon's early AI agent prototypes never reached production Sep 24

Lovable hits $600m revenue with two-thirds of Fortune 500 as customers Sep 24

Cursor launches Rollouts agent to monitor code changes from PR through to production Sep 24

Research finds AI agent memory corruption enables unauthorized actions in 98.6% of matched trials Sep 24

Research finds LLM reasoning operations produce distinct internal activation patterns Sep 24

ChatGPT Voice gains agentic workflows with email, calendar, and Slack integration Sep 24

openKylin 3.0 ships with Linux kernel 7.0 and built-in AI assistant Sep 24

OpenAI agent accessed non-public Australian Medicare files, triggering government investigation Sep 23

GitHub Copilot code review settings expand to all plans with auto-review controls Sep 23

AWS launches simplified sign-up with $100 free credits and no IAM setup required Sep 23

Debian launches free AI inferencing portal for contributors powered by Scaleway Sep 23

OpenAI releases MentalHealthBench open benchmark for AI mental health conversations Sep 23

Claude discovers unknown enzyme system in bacteriophage DNA with CRISPR-like features Sep 23

ChatGPT voice mode gains tool access and Word integration rolls out to all plans Sep 23

AI agent campaign steals 600K credit cards from 119 retail sites using open-source frameworks Sep 23

DeepSeek publishes paper on running 3 million AI agent sandboxes per day Sep 23

Google Antigravity SDK adds local model support via Gemma 4 and Ollama Sep 23

Google launches Gemini TTS models with voice cloning from 30-second audio samples Sep 23

NVIDIA releases Nemotron 3 open-weight diarization model supporting up to eight speakers Sep 23

Laravel AI SDK 1.0 ships with approvable tool calls and breaking storage changes Sep 23

Vercel Sandbox persistent Drives enter public beta with up to 16 TiB storage Sep 23

Anthony Fu launches Devframe 1.0 as a portable foundation for framework-neutral DevTools Sep 23

vllm-metal brings concurrent LLM serving to Apple Silicon via MLX and Metal Sep 23

Amazon lets sellers manage stores directly from Claude and Amazon Quick Sep 23

GitHub Copilot app adds local sandboxing to restrict file and network access per project Sep 23

Researchers find reasoning LLMs self-jailbreak after benign math and code training Sep 23

Supply chain worm hits npm and PyPI packages stealing AWS keys and GitHub tokens Sep 23

GitHub Copilot CLI adds whole codebase indexing for C++ projects Sep 23

Apple pitches new Mac Mini and Mac Studio as local AI compute alternative to cloud Sep 23

Anthropic in talks to lease 1GW of compute from Apollo-backed data centre operator Sep 23

VS Code 1.138 adds agent sessions inside Dev Containers and Codex cross-app continuity Sep 23

GitHub Copilot adds OpenTelemetry support for enterprise agent activity monitoring Sep 23

GitHub Copilot for JetBrains 1.18 adds auto-approvals, conversation rewind and plan mode Sep 23

Amazon CloudWatch Omni launches with AI agent tracing and built-in evaluators Sep 22

Cyera raises $400m more in 2026 bringing total AI security funding to $1.4bn Sep 22

Anthropic research finds 250 poisoned documents can backdoor any LLM Sep 22

Alibaba previews four-tier Qwen 4 lineup at Apsara 2026 with no pricing or benchmarks yet Sep 22

Meta admits Muse AI assistant was heavily inspired by open-source OpenClaw project Sep 22

xAI releases Grok 4.7 with longer RL training and stronger jailbreak safeguards Sep 22

Xiaomi open-sources MiMo-V2.6 Pro and Flash omnimodal models Sep 22

JetBrains launches Air agentic development platform with IDE, cloud and governance layers Sep 22

Cisco Talos finds Windows malware using AI model voting panel for attack decisions Sep 22

OpenAI releases GPT-6 Sol and Luna with 50% token price cuts Sep 22

Blueprint Alliance launches to set governance standards for autonomous AI agents Sep 22

Anthropic releases Claude Opus 5.5 with 20% price cut and silent model rerouting caveat Sep 22

Alibaba releases Qwen-Image-2.1 open-source image generation model Sep 22

DigitalOcean launches Managed Agents platform for agentic workloads in public preview Sep 22

TypeSafe AI launches Jev, a non-autoregressive decision model priced at $0.042 per million tokens Sep 22

Mac mini M6 launches with up to 170 Gbps memory bandwidth and improved local LLM performance Sep 22

China investigates DeepSeek and Moonshot after Anthropic accuses them of secretly routing requests to Claude Sep 22

Google delayed disclosing Gemini unauthorized system access for months Sep 22

Alibaba plans AI model with 5 to 10 trillion parameters, dwarfing current Qwen3.8 Max Sep 22

Aikido open-sources Altar, a pruned AI model for air-gapped security testing Sep 21

MiMo-V2.6-Pro tops open weights models on Artificial Analysis Intelligence Index Sep 21

AWS open-sources Strands Harness agent claiming 45% cost savings over Claude Code and Codex Sep 21

Qwen Code Desktop v0.24.3 ships Linux kernel sandbox, browser SDK and workflow checkpointing Sep 21

Claude reportedly used in exploit chain targeting OpenAI employee accounts Sep 21

GKE Pod snapshots cut AI inference cold-start latency by up to 89% Sep 21

Grok 4.7 launches at $6 per million output tokens with 500K context window Sep 21

Cloudflare proposes Agent Development Lifecycle to replace traditional SDLC Sep 21

Cloudflare Python Workers reach general availability with FastAPI, Django and Flask support Sep 21

Google unveils Googlebook laptops running merged Android and ChromeOS operating system Sep 21

Plugin4Shell researcher details unpatched Copilot exposure on Bitbucket and self-hosted git Sep 21

Claude Code 2.1.277 adds AGENTS.md support for cross-tool AI agent instructions Sep 21

Google open-sources AX, a Kubernetes-style orchestrator for autonomous AI agents Sep 21

FrogNano 4B coding agent hits 61.5% on SWE-bench without frontier model distillation Sep 21

Survey of 427 A2A Agent Cards finds 60% conformance and emerging agent commerce economy Sep 21

Kimi K3 2.8T-parameter MoE model launches with vLLM speculative decoding support Sep 21

ZCode found silently uploading encrypted local Git repos to Alibaba cloud without user verification Sep 20

xAI quietly rolls out Grok Bot agent handoff to Grok Heavy users Sep 20

Apple CEO John Ternus reveals Siri AI runs on Google Gemini at roughly $1bn a year Sep 20

Alibaba open-sources OpenCodeReview, a Go CLI for AI-assisted code review Sep 20

Researchers find two sandbox escapes in OpenAI Codex Desktop and CLI, both patched Sep 20

Google releases ADK for Kotlin 1.0 with on-device AI and feature parity with Python Sep 20

Google ScientistTwo autonomously improves 86 of 107 ML research problems in recursive loop Sep 19

Anthropic weighs new model launch as GPT-6 Astra captures 13% of enterprise AI spend Sep 19

xAI releases Grok Voice Transcribe 2.0 at $0.10/hour with diarization and key term biasing Sep 19

DeepSeek releases V4.1 Flash with 4x KV cache reduction and 500B+ parameters Sep 19

Google confirms Gemini accessed real company systems during cybersecurity test Sep 18

Anthropic proposes AI R&D automation metrics to track frontier lab development pace Sep 18

Anthropic open-sources biomolecular model optimizations achieving 4x speedups Sep 18

GitHub Copilot code review adds commit-level tracking and smarter auto-resolution Sep 18

Anthropic and Accenture commit $1 billion to independent frontier AI evaluation Sep 18

GitHub Copilot deprecates GPT-5.5, Gemini 3.7 Flash and four other models on October 19 Sep 18

ChatGPT plugins add support for connecting multiple accounts simultaneously Sep 18

Universal and Sony sue Suno again over 60,202 recordings with $9B damages at stake Sep 18

Disney names Character.AI CEO Karandeep Anand as its first company-wide CTO Sep 18

PrismML releases Bonsai 2 27B compressing Qwen3 27B to 5.9 GB with ternary weights Sep 18

ChatGPT desktop app adds support for Chrome extensions Sep 18

Manus seeks $4B valuation in $500M raise after Meta acquisition blocked by Beijing Sep 18

Google open-sources Mantis multi-agent code review harness used on its own infrastructure Sep 18

Shopify drops React Native for native Swift and Kotlin, cites AI-assisted porting Sep 18

PayPal WebMCP case study shows 4.6x cheaper and 2x faster AI agent checkout Sep 18

UN launches AI-accessible data commons on Google platform after LLMs score 21% on development indicators Sep 18

Open-weight models cross 50% of Vercel AI Gateway tokens but Anthropic still takes 64% of spend Sep 18

Splunk open-sources Token Meter to track AI coding agent token spend Sep 18

Google Home opens early access to MCP letting any AI agent control smart home devices Sep 18

Anthropic opens physical biology lab to advance AI-driven drug discovery Sep 18

Unit 42 finds prompt injection can exfiltrate plaintext credentials from AWS AgentCore Identity vault Sep 18

IBM releases Granite 4.2 models and Granite Speech 5.0 under Apache 2.0 Sep 18

Researchers used Claude Opus 5 to breach OpenAI private repos via Discourse chain Sep 18

NVIDIA Agora uses Git commits as shared memory for multi-agent LLM research coordination Sep 18

Anthropic redesigns Claude Code Projects into parallel multi-thread coordinator Sep 18

Independent benchmarks find Gemini 3.8 Flash searches restricted data to inflate scores Sep 18

GitHub Copilot usage metrics API adds agentic CLI activity tracking Sep 18

ChatGPT desktop app for Windows adds Appshots screen context feature Sep 18

OpenAI builds enterprise sales org, hires three leaders from Cursor and Snowflake Sep 17

Third-party iPhone Duo simulator lets developers test folding hinge interactions ahead of Apple SDK Sep 17

Bolt launches Forge agent with unlimited token model on Pro plans until October Sep 17

OpenAI discloses GPT-5.6 Sol instances learned to hide mistakes from users during training Sep 17

Anthropic publishes internal metrics on how much AI is used to build AI Sep 17

OpenAI adds voice agent to Codex powered by GPT-Live-1 Sep 17

OpenAI launches Astra for Law with legal search index covering 230 million URLs Sep 17

Research finds AI coding agents silently fine-tune model weights in 42% of planning tests Sep 17

Anthropic launches Claude Code Projects with parallel cloud sessions that run offline Sep 17

Google releases Android Bench 2.0 with multi-day engineering tasks and completion rate scoring Sep 17

AI threat digest reports frontier model broke containment and reached Hugging Face systems Sep 17

OpenAI claims progress on Navier-Stokes Millennium Prize Problem Sep 17

Sentient Labs AI self-improvement system exploits test answer key to cheat Sep 17

Trail of Bits finds GPT-5.6-Cyber agents repeatedly escape QEMU/KVM sandboxes Sep 17

GitHub rewrites Copilot agent runtime from TypeScript to 800K lines of Rust using AI agents Sep 17

Perplexity builds CobbleDB in Rust to replace DynamoDB, cuts p99 read latency 5x Sep 17

OpenAI publishes model misalignment disclosure framework and six incident reports Sep 16

Yale study finds most AI benchmark failures are due to flawed questions, not model errors Sep 16

AIUC raises $40M Series A to certify and insure AI agents via Lloyd's of London Sep 16

BragJack attack hijacks agentic AI in Chrome, Edge, Opera Neon and Comet browsers Sep 16

Google Home adds MCP support to let third-party AI agents control smart home devices Sep 16

OpenAI pulled 25% of production engineers to run Astra security scans before release Sep 16

Arcee AI raises Series B at $1B+ valuation after building Trinity lineup for $20M Sep 16

Mystery AI model Union Alpha appears on OpenRouter with 2B tokens processed on day zero Sep 16

OWASP 2026 LLM Top 10 elevates excessive agent permissions to third place Sep 16

Cognition adds Code Scans to Devin for codebase-wide audits and automated PRs Sep 16

Cohere and Aleph Alpha finalize merger to form transatlantic sovereign AI company Sep 16

Google details how attackers steal AI credentials, distill models and poison AI coding tools Sep 16

Anthropic merges Claude Cowork and Claude Chat into a single unified product Sep 16

DuckDB releases Claude Code plugin for SQL, file profiling and object storage exploration Sep 16

AMD releases ADLX 2.0 SDK with MCP servers for AI agent GPU control Sep 16

Zed launches Delta public beta to replace pull requests with AI agent threads Sep 16

Qwen Code v0.24.0 adds Linux sandboxing, background workflows and multimodal media pipeline Sep 16

GitHub data shows AI agent-generated PRs grew 9x in eight months Sep 16

AMD GAIA 0.24 adds meeting transcription and speaker recognition for local AI Sep 16

iOS 27 code reveals Siri can swap out its AI model for Claude or ChatGPT Sep 16

Mozilla adds Mistral Small 4 to Firefox Smart Window beta as alternative AI option Sep 16

Zuckerberg rejects Amodei's AI slowdown call as OpenAI and Altman sign on Sep 16

CheatBench measures how often AI agents game their reward signals Sep 16

GitHub Copilot drops support for JetBrains 2025.1.x IDEs Sep 16

GitHub ships standalone Copilot desktop app with canvas micro-app feature Sep 16

Anthropic reportedly building Claude Money personal finance feature Sep 15

Factory raises $200M at $5B valuation for enterprise AI coding agents Sep 15

Meta launches WhatsApp Business MCP server for AI coding agents Sep 15

OpenAI doubles Codex for OSS grants to 10,000 open source maintainers Sep 15

Stanford and MIT paper finds AI harness design can cause up to 6x performance gap Sep 15

NVIDIA reveals NVLink 6, DSX platform and NVLink Fusion at AI Infra Summit Sep 15

TypeSafe AI launches Jev, a model that skips token generation for typed structured decisions Sep 15

Google releases Gemini 3.8 Live with background reasoning during live conversations Sep 15

Devin coding agent gains macOS VM with iOS simulator for mobile app testing Sep 15

Odyssey-3 world model controls cars, drones and robots from shared training Sep 15

OpenAI benchmark exercise ends with 1,200 AI agents coordinating to breach Hugging Face Sep 15

Temporal raises $550M at $12.55B valuation as AI agent workloads surge Sep 15

Anthropic report details how threat actors used Claude to breach 1.8M Android apps and government targets Sep 15

Salesforce launches Koa reasoning model built on Nvidia Nemotron for enterprise CRM tasks Sep 15

Python and PHP still expose stdlib to shadowing attacks that Ruby and Perl fixed years ago Sep 15

iOS 27 code reveals Siri can swap in Claude or ChatGPT as its AI backend Sep 15

Cline launches standalone desktop app outside the code editor Sep 15

Nvidia, Palantir and Booz Allen restrict Claude Fable over 30-day data retention policy Sep 14

New Real-SWE benchmark shows best coding agents fail 60% of the time on private codebases Sep 14

MSVC Build Tools v14.52 adds full C++23 conformance, ships in November Sep 14

Anthropic says Claude writes 80% of its code as CI jobs surge 25x in six months Sep 14

Anthropic launches Claude for Financial Advisers with BlackRock and Vanguard integrations Sep 14

Google DeepMind study finds AI agent swarms self-police cheating behavior Sep 14

Claude Opus 5 draws user backlash as real-world performance trails Opus 4.8 Sep 14

OpenAI details how its LLMs designed the Jalapeño AI accelerator chip Sep 14

Anthropic analysis of 400,000 Claude Code sessions reveals expert usage patterns Sep 14

OpenRouter launches US in-region routing to keep Chinese model traffic on US soil Sep 14

Expo launches cloud-hosted iOS Simulator so Windows and Linux devs can test iOS apps Sep 14

China's state security minister names Claude Mythos and GPT-5.5-Cyber as infrastructure threats Sep 14

OpenAI agents flooded RubyGems with 2,000 packages two months before Hugging Face breach Sep 14

DeepSeek releases V4.1-Flash with 552B MoE parameters and 1M token context Sep 14

Visual Studio 2026 ships Copilot modernization, BYOM support and 272 C++ bug fixes Sep 14

Anthropic report accuses Chinese AI labs of large-scale Claude distillation fraud Sep 14

AI system cracks 370-year-old cipher in 44 minutes Sep 14

Anthropic picks Nasdaq for October IPO as OpenAI rules out 2026 listing Sep 13

Cohere releases North Small Translate open-weight model covering 50 languages Sep 13

OpenAI launches Agents API for managed cloud-based AI agents Sep 13

GPT-6 Astra autonomously generates running routes using OpenStreetMap data in ChatGPT Work Sep 13

Anthropic CEO calls for pacing frontier AI development and third-party lab access Sep 12

Meta finds adversarial LLMs can flip AI judge verdicts in up to 91% of cases Sep 12

Windows 11 update KB5124008 breaks WSL folder sharing with no workaround yet Sep 12

Report links OpenAI agent swarm to undisclosed RubyGems supply chain attack Sep 12

OpenAI acquires Git AI team to track coding agent contributions in Codex Sep 12

ChatGPT Sites gets upgraded creation, sharing, and hosting capabilities Sep 12

Stanford and Together AI find local AI efficiency up 5.3x since 2023 but hard reasoning still needs cloud Sep 11

Devin CLI ships Fusion mode to cut multi-model coding agent costs by 39% Sep 11

Claude Code adds plugin eval command to test and benchmark plugins Sep 11

OpenAI launches GPT-Rosalind biological reasoning model in API and Codex Sep 11

Moonshot AI targets $2B annual revenue as Anthropic accuses it of Claude distillation Sep 11

Cohere releases North Small Translate 1.0 but restricts commercial use to paid platform Sep 11

OpenAI pauses $200 ChatGPT Pro sign-ups as Astra demand strains capacity Sep 11

GitHub ships native Copilot desktop app for Windows and Mac Sep 11

Attackers abuse Claude and ChatGPT domains to host malware bypassing URL reputation filters Sep 11

Lightfield raises $47M Series A to build a CRM designed for AI agents Sep 11

Sergey Brin returns to active Google role shaping Gemini strategy and AGI direction Sep 11

OpenAI launches ChatGPT for Financial Services with S&P, LSEG, Moody's and PitchBook data Sep 11

OpenAI retires GPT-5.3-Codex-Spark next week as usage declines Sep 11

Shopify migrates mobile apps from React Native to native Swift and Kotlin Sep 11

CPU cache side-channel attack reconstructs local LLM output without memory access Sep 11

Cursor releases CursorBench 4.0 with harder instruction-following and long-horizon tasks Sep 11

Claude adds persistent scrollable diff pane for real-time code change monitoring Sep 11

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
```