Adobe's EditAgent for Photoshop
A new system that selects specialized professional tools rather than relying on a single general-purpose editor.
🤖 AI — 2026-11-01
ddNews · a daily AI briefing for Dirk (generative video/ddStudio, Lit
Database bottlenecks for AI agents
Enterprises find that agentic architectures require databases capable of handling thousands of unplanned, millisecond-latency queries.
Adobe's EditAgent for Photoshop
Research with UC San Diego shows an agent capable of selecting professional tools for complex image editing tasks.
Vision model optimization
New findings suggest that adding labels to images is more effective for model reasoning than prompt engineering.
OpenAI cracks down on account sharing
New limits on paid accounts per device and payment card aim to stop "subscription hopping."
Google's StreamVibe video agent
A new research project aims to create video AI that recognizes unspoken user needs during physical tasks like cooking.
Google's StreamVibe research
Researchers are developing video agents capable of recognizing unspoken user needs during physical tasks like cooking.
OpenAI merges Chat and Work modes
Starting December, Plus, Pro, and Business plans will move to a shared, token-based quota for ChatGPT and Codex.
OpenAI merges Chat and Work modes
Starting December, Plus, Pro, and Business plans will move to a shared, token-based quota for ChatGPT and Codex.
AWS capacity sold out through 2027
CEO Andy Jassy confirms large committed cloud capacity is nearly full, highlighting extreme demand for AI compute.
Claude Code game restrictions
Anthropic updated Opus 5.5 instructions to prevent the recreation of existing games while allowing original development.
Gemini 4 Carbon
Google's new model variant specifically optimized for complex coding and cybersecurity.
Cloudflare Post-Quantum GeoKey Manager beta
A rebuild using hybrid ML-KEM-768 and X25519 to protect geo-fenced keys.
Anthropic expands watermarking for EU AI Act
New watermarks are being rolled out to meet December deadlines, though loopholes exist in Claude 4.6 models.
EU AI Act drives Anthropic watermarking
New text watermarks are being rolled out to meet December deadlines, though loopholes exist in older models.
OpenAI cybersecurity training
Plans to use Daybreak Blue session data for training, sparking privacy concerns among security researchers.
Claude blocks game reverse-engineering
Anthropic has implemented broad restrictions to prevent the model from analyzing video game code.
Adversarial training data
Anthropic demonstrated that seemingly harmless data (e.g., beekeeping tips) can train smaller models to sabotage oversight.
Meta Muse Charm hardware
Meta is launching a "Tamagotchi-style" wearable for its AI helper to increase consumer engagement.
Volta-Autodoxer AI malware
A new infostealer uses AI to automatically collect and organize victim data.
Microsoft pivots workforce for AI
The company is offering voluntary severance to 60,000 employees to redirect billions into AI infrastructure.
Meta's Muse impact
The AI agent is positioned as a potential disruptor for various business sectors.
Microsoft workforce shift
Offering voluntary severance to 60,000 employees to redirect billions toward AI infrastructure.
DeepMind's ForTy global forest map
High-resolution (10m) mapping of primary forests and plantations for EU deforestation compliance.
AI agent performs brain injection
A biotech startup successfully used an agent to select a gene therapy and inject it into a live mouse.
OpenAI funding shortfall
OpenAI raised only $11B of a targeted $30B, suggesting a cooling of investor appetite despite a $1.25T valuation.
OpenAI funding shortfall
The company raised only $11B of a targeted $30B, signaling a potential cooling of investor appetite despite a $1.25T valuation.
Kubernetes Confidential Containers integrate Fortanix
New attestation and key management for secure GPU-verified AI workloads.
Google's PageBreak AI finds 500+ XSS bugs
An AI security agent using Gemini models successfully uncovered hundreds of vulnerabilities.
Industry-wide AI crawling opt-out
Google, Microsoft, OpenAI, and Anthropic are adopting a new standard for publishers to block AI scraping.
Nvidia considers cosigning Nscale loans
The chipmaker may act as a cosigner for AI data center loans in West Virginia to stabilize investor confidence.
Dulo funding
Sebastian Thrun's AI engineering startup raised $130M.
Intel stock crash
Shares plummeted due to weak PC market guidance as AI demand shifts components away from consumer hardware.
Anthropic faces 'junk' credit rating
Reports suggest high borrowing costs ahead of the IPO due to the company's current negative cash flow.
$6.3B Binance heist
The largest crypto theft in history occurred, dwarfing previous records.
Shift to "Systems Race"
The semiconductor industry is moving from competing on individual chips to treating them as components of larger systems.
Extreme Networks pushes on-prem AI
Preview of "Exchange zones" that utilize spare access point and switch compute for local AI.
Claude autonomously operates on mouse brain
An AI agent independently selected gene therapy and coordinates to perform surgery on an epileptic mouse.
Claude autonomously operates on mouse brain
Claude Sonnet 5 independently selected medication and coordinates to perform a successful brain surgery.
Apple Intelligence arrives in China
Features will launch with iOS 27.3, starting with a beta release.
Malicious Claude Code extension in Cursor
A tampered extension affected 5,000 users, attempting remote access and data theft.
Major AI labs adopt scraping opt-out standard
OpenAI, Google, Microsoft, and Anthropic agree to a new standard for publishers to block AI crawlers.
Ludo Robotics Ludi 0.2
A new robot intelligence model designed specifically to understand and interact with human intent.
OpenAI updates Model Spec
New behavioral guidelines focus on "faithfulness to human intent" and using conversation history to derive user goals.
OpenAI updates Model Spec
New behavioral guidelines focus on five core values, including faithfulness to human intent and using conversation history to derive user goals.
Reddit warns of Google AI traffic loss
CEO Steve Huffman reports a loss of over 1M search referrals per week due to AI Overviews.
Google's first TPU satellite is online
A Gemma 4 model is now running on-orbit, marking the first TPU deployment in space.
Anthropic files for IPO
The filing lists 88 risk factors across 93 pages, signaling a massive but risky market entry.
Anthropic warns of diseconomies of scale
The IPO prospectus highlights 88 risk factors, including the massive spending required to maintain its competitive edge.
Google's space-borne TPU
The first AI server in orbit has successfully booted a Gemma 4 model and transmitted responses to Earth.
DoorDash cuts latency via batch LLM inference
Grocery reorder times dropped from 16s to 2s using precomputed recurring needs.
Runware building prefabricated AI data centers
A new approach to mass-producing data center infrastructure on an assembly line.
RBAC becomes the AI agent bottleneck
Enterprises find that role-based access control, not model capability, is the primary hurdle for production agents.
Data platform accountability
The focus is shifting from AI pilots to measurable business returns, putting pressure on data sovereignty and cost.
Netflix treats GPU reliability as distributed systems problem
Detailed implementation of layered health checks and automated remediation for large-scale training.
Binance heist
A $6.3B crypto theft, potentially by North Korean hackers, marks the largest digital heist in history.
Custom hooks for Claude Code
Implementation of mods to protect API keys and monitor context usage with 100x less overhead.
Dream launches AI architecture tool
The startup raised $17M to generate professional building designs using generative AI.
Intel stock crash
Shares plummeted over 30% due to weak guidance and a slowdown in the PC market.