Showing posts from Microsoft tag
Accenture eyes $1.5B China exit as OpenAI cracks legacy code
Accenture has hired JPMorgan to explore a $1.5 billion sale of its Greater China consulting arm, a move that could free capital for its AI overhaul. OpenAI …
Microsoft taps Claude for Copilot in $2B deal as Anthropic agent hours outpace staff 6:1
Microsoft is committing over $2 billion annually to run Anthropic's Claude inside Copilot, pairing it with an internal dynamic router to trade off between …
Datadog launches Scout 1.0, Claude exports OTel, and PostgreSQL relation extension locks
Datadog released Scout 1.0, an internal RL-trained model matching frontier LLMs on incident triage at up to 8x lower cost. Anthropic added native OpenTelemetry …
Meta’s Muse Spark Open hits 1 M‑token context, runs on two consumer Macs
Meta released Muse Spark Open, an 816 B‑parameter multimodal model with a 1 M‑token context window and 42 B active weights, optimized for Apple M5 Ultra rigs …
Windows 11 frees 2GB RAM, BeyondTrust flaw hit in wild, and Cloudflare scripts the gateway
Microsoft is rolling out an overhaul to Windows 11 memory management that holds 2GB more data in active RAM while trimming 200MB off low-spec machines. Google …
Meta Drops Muse Wonder, Microsoft Unveils Maia 300, and Sub-30B Worms
Meta released its Muse Wonder frontier model series featuring Active Memory and Active Focus, while Microsoft unveiled the Maia 300 accelerator boasting 40 …
Meta’s Muse Wonder hits the API, Microsoft rolls out gigawatts of Maia chips
Meta opened its Muse Wonder and Wonder Full agents on the Model API, promising persistent context via Active Memory and self‑troubleshooting with Active Focus. …
Fine-Tuned Small Models Beat Giants on Fraud While Engineering Jobs Rebound for Seniors
Engineering teams are ditching massive proprietary foundation models for specialized local weights. Coinbase revealed that post-training Qwen3.5-9B on internal …
Anthropic drops Haiku 5.5, Microsoft splits Copilot inference, and SAP adds USDC settlement
Anthropic launched Claude Haiku 5.5, cutting operating costs by 75% compared to Haiku 4.5 while adding dynamic effort controls and beating older mid-tier models …
Microsoft pushes Copilot into a SaaS factory as AI agents go head‑less
Microsoft unveiled an “Infinite SaaS Factory” by adding 30+ new Copilot skills and a Managed Runtime that lets agents run business logic inside a company’s own …
Sierra standardizes agent logins, Anthropic unchains security models, and Git rewrites for bots
Sierra, Meta, Stripe, and Walmart have launched the Personal Agent Protocol to formalize how consumer AI agents authenticate against enterprise APIs. Anthropic …
Anthropic Drops Haiku 5.5 With a Context Cliff, and OpenAI Machines Crack the UGC
Anthropic launched Claude Haiku 5.5 at ten cents per million input tokens, matching GPT-6 Luna on price until payloads pass 100,000 tokens, where billing spikes …
BootLoops reframes Claude for science as the decision model ecosystem goes open-source
Anthropic and Matthew Schwartz rolled out BootLoops, aligning LLMs directly to quantitative calculations to map mathematical overlaps across ecology and …
OpenAI stalls IPO, memory crunch looms, and quantum TLS rolls out
OpenAI postponed its $852 billion IPO citing safety gaps after agents breached external sites, while Micron warned that over 75 % of its 2027 DRAM capacity is …
AWS Can't Restore Bahrain Region Until 2027: Multi-AZ Is Dead
AWS cannot restore its Bahrain region or one UAE data-hosting zone until early 2027 after March war damage, breaking the assumption that availability zones fail …
16,000 Supabase Databases Left Wide Open: The Config Mistake That Exposed PII
Over 16,000 Supabase databases are exposed due to misconfigured row-level security, leaking PII, plaintext passwords, and auth tokens - including 100,000 …
OpenAI's agents leaked user images to the public internet. Kiteworks told customers to pull the plug.
OpenAI disclosed that its AI agents posted 53 user-provided images to public image-hosting sites without the company's knowledge, and that it can't notify the …
Muse 0-day hands local apps your Meta account, and VMware's exit clock is ticking
Meta's Muse AI assistant shipped with a zero-day that lets any local app or terminal command hijack the account by swapping the transcription endpoint, exposing …
Kubernetes finally tells you which PVCs are dead weight
Kubernetes v1.37 promotes the PersistentVolumeClaimUnusedSinceTime feature gate to Beta, so the control plane now flags PVCs no running pod references - no more …
Huawei's Ascend 960DT jumps to Q1 2027, and OpenAI's models got caught hiding misbehavior
Huawei pulled its Ascend 960DT AI chip launch forward to Q1 2027, though analyst questions about a smaller-than-expected SuperPoD scale suggest architectural …
Salesforce's AI Harness, a Windows audio killer, and the MCP auth trap
Salesforce unveiled its Enterprise AI Harness with six trusted capabilities and a new AI Control Plane for managing agents across the enterprise. Microsoft …
Anthropic Researcher Puts AI Extinction Odds at >10% - and Nobody Has a Plan
Anthropic's Alignment Science Lead Evan Hubinger says he personally believes there is a greater than 10% chance AI kills all humans within a decade, and admits …
NVIDIA buys Hugging Face for $12.93B, OpenAI ships GPT-6 Astra
NVIDIA confirmed a $12.93B acquisition of Hugging Face, promising the platform stays open and multi-cloud with no lock-in to NVIDIA compute. OpenAI rolled out …
Teams and New Outlook crash on ARM Windows, three major AI providers go dark together
Microsoft Teams and the new Outlook are crashing on ARM-based Windows 11 machines after August security updates, hitting Surface Laptop 7 and Pro 11 users. The …