New model
IMPACT 54FLUX.1 — Black Forest Labs
High-fidelity image generation with strong text rendering, usable for layout and brand mockups.
What is happening right now
Every event below is verified, classified and scored by our internal analysis pipeline, then mapped to the careers it affects. No raw feeds, no unvetted links - just the verdict.
Feed
Disruption feed (30 events)
Our verdicts, newest first - with disruption scores and affected careers. Underlying research and data collection methods stay proprietary.
- New modelIMPACT 46
Open-weight frontier-class models enabled self-hosted automation
Llama 3.1 405B and subsequent open-weight releases let organisations run near-frontier models on their own infrastructure, changing the cost and compliance barriers to automation.
- New modelIMPACT 50
Llama 3.1 405B — Meta
Open weights enabled widespread self-hosted deployment of general-purpose assistants.
- AI adoption signal
AI exposure is highest in high-skill occupations, with polarised effects
The OECD Employment Outlook reports that occupations most exposed to AI are predominantly high-skill, and that exposure does not equal job loss, while noting emerging evidence of effects on task composition and entry-level work.
- New modelIMPACT 54
Claude 3.5 Sonnet — Anthropic
Strong document reasoning, computer use (later revisions) and code generation.
- New AI toolIMPACT 42
Runway Gen-3 Alpha — Runway
Text/image-to-video, motion brush and camera controls.
- New modelIMPACT 55
GPT-4o — OpenAI
Real-time voice conversation, vision understanding and high-quality document reasoning.
- New modelIMPACT 62
GPT-4o introduced real-time speech-to-speech conversation
GPT-4o added a single model handling text, audio and image, enabling sub-second conversational voice — the capability precondition for substituting routine telephone work.
- Research publishedIMPACT 45
Macroeconomic gains from AI are likely modest over a decade
Estimating that only a small share of tasks are currently cost-effective to automate, the paper projects total-factor-productivity gains of no more than about 0.66% over ten years, with limited aggregate labour-demand effects.
- New modelIMPACT 53
AlphaFold 3 — Google DeepMind
Structure prediction across biomolecular interactions, used in drug-discovery research pipelines.
- Labour market updateIMPACT 62
Automation-prone freelance work saw fewer projects and lower earnings after ChatGPT
Using platform data from a large online labour market, the authors find that after ChatGPT's release, jobs more exposed to AI automation experienced a relative decline in the number of projects and in freelancer earnings compared with less-exposed jobs.
- Labour market updateIMPACT 35
Companies That Have Replaced Workers with AI in 2025 and 2026
This is demonstrated in our annual ... that use artificial intelligence extensively. The result of this adoption? Widespread job insecurity. While it is true most companies are planning to augment human tasks rather than replace them altogether, the actions some companies are taking are confirming commonly held fears about the direction of the employment landscape. Understandably, most companies don’t want to gain a reputation for replacing workers with ...
- AI adoption signalIMPACT 42
AI may expand expert work rather than only substitute for it
The authors argue that AI can enable workers with less formal training to perform expert tasks, potentially rebuilding middle-class occupational demand rather than only displacing it.
- Research publishedIMPACT 56
About 40% of global employment is exposed to AI, 60% in advanced economies
Using an AI exposure index, the IMF estimates that around 40% of global employment and around 60% of employment in advanced economies is exposed to AI, with exposure concentrated in high-skill occupations.
- New modelIMPACT 46
Mobile ALOHA — Stanford University
Demonstrated generalisable household-style manipulation tasks, a marker for physical-task automation progress.
- New capabilityIMPACT 34
Low-cost imitation learning for generalisable mobile manipulation
Mobile ALOHA demonstrated bimanual mobile manipulation on everyday tasks using imitation learning on relatively low-cost hardware.
- New AI toolIMPACT 43
Sierra — Sierra
Voice and chat agents that execute account actions, not just answer questions.
- New AI toolIMPACT 41
Descript — Descript
Text-based editing, studio sound and overdub voice replacement.
- New modelIMPACT 88
Whisper large-v3 made accurate transcription freely self-hostable
Whisper large-v3 provides open-weight multilingual speech recognition, allowing transcription to be run at scale for the cost of compute alone.
- New modelIMPACT 55
Whisper large-v3 — OpenAI
Cheap, accurate transcription at scale — the direct technical basis for replacing manual transcription.
- Research publishedIMPACT 20
Dario Amodei's remarks at the AI Safety Summit
Anthropic's CEO presents the Responsible Scaling Policy and its AI Safety Levels framework at the UK AI Safety Summit.
- Research publishedIMPACT 20
Expanding access to safer AI with Amazon
Amazon will invest up to $4 billion in Anthropic and become its primary cloud provider, with models trained on AWS Trainium and Inferentia.
- Research publishedIMPACT 20
Prompting Claude's long context window
Two techniques improve Claude's long-document Q&A: extracting relevant quotes first and providing in-context examples.
- New modelIMPACT 45
Introducing Anthropic's Responsible Scaling Policy
Our Responsible Scaling Policy is a series of technical and organizational protocols to manage the risks of increasingly capable AI systems.
- Research publishedIMPACT 20
The Long-Term Benefit Trust
An independent body of five trustees with authority to elect and remove Anthropic board members, balancing stockholder and public interests.
- Research published
Consultants performed better inside the AI frontier and worse outside it
In a field experiment with 758 consultants, GPT-4 users completed about 12% more tasks, about 25% faster, with about 40% higher quality on tasks within the model's capability frontier — but were about 19 percentage points more likely to produce incorrect answers on tasks outside that frontier.
- Research publishedIMPACT 20
Anthropic partners with BCG
Anthropic and Boston Consulting Group partner to bring Claude to BCG's enterprise clients for knowledge management and data analysis.
- New modelIMPACT 45
Introducing Claude Pro
Claude Pro is a subscription for claude.ai offering 5x more usage, priority access during peak periods, and early access to new features.
- Research publishedIMPACT 74
LLMs affect at least 10% of tasks for about 80% of US workers
Using O*NET task descriptions scored by GPT-4, the authors estimate that about 80% of the US workforce could have at least 10% of their tasks affected by large language models, and about 19% could see at least 50% of their tasks affected.
- Research publishedIMPACT 20
SKT partnership announcement
SK Telecom invests $100 million in Anthropic and partners to build a telco-tuned version of Claude for customer service, marketing, and sales.
The Live Disruption Radar streams continuously. Each entry shows our independent assessment - impact score, confidence and affected careers.