GPT 5.6 keeps Microsoft tied to OpenAI while breakup talk grows
OpenAI says GPT 5.6 remains the preferred model for Microsoft Copilot 365. The practical question is not partnership theater, but who supplies the brain of enterprise office AI.
Lilith · selected stories
News
What is actually happening in AI. Selected stories, context and opinion without the promotional noise.
Atom feed ↗Codex running for hours turns chat into delegated work
Ars Technica says OpenAI has reframed Codex around independent workflows that can run for hours. For teams, the hard part is not code generation, but supervising an agent while it works.
Claude Wrapped turns chatbot use into a mirror of habits
Anthropic is adding a reflect feature to Claude that shows topics, task types and usage patterns across one month, 3 months, 6 months or a year. User analytics is becoming part of the human AI relationship.
GPT-5.6 gets the green light as ChatGPT Work pushes Codex beyond developers
OpenAI is publicly rolling out GPT-5.6 after a limited preview for government approved organizations and introducing ChatGPT Work. The real shift is Codex style capability moving into documents, spreadsheets, presentations and everyday business workflows.
Claude Reflect sells AI by showing users their own dependence
Anthropic is rolling out Reflect, a Claude dashboard for topics, usage patterns and AI habits. It looks like personal analytics, but it also reminds users how much of their work now flows through the chatbot.
Anthropic talks to Samsung about its own chip because depending on Nvidia costs too much
According to The Information, Anthropic is discussing a custom chip collaboration with Samsung, a week after OpenAI and Broadcom unveiled the Jalapeño inference processor. Custom silicon is turning into standard equipment for frontier labs.
Fanfiction found an AI detector that raises alarms faster than it proves authorship
The Verge describes a fight around an AO3 skin that turns pages red when text pasted directly from Claude carries the font-claude-response-body artifact. The signal is technically interesting, but it has quickly become a social trial of authors even though it shows a copy path, not how much of a work was written by AI.
General Intuition bets $320 million that game data can train physical AI
TechCrunch profiles General Intuition as a startup using game data to train world models for robotics and physical AI. The numbers are attention grabbing, but the real question is whether in-game actions transfer into useful intuition in the real world.
GPT-Live-1 tackles voice AI’s hardest habit: knowing when to shut up
OpenAI is rolling out GPT-Live-1 for ChatGPT voice mode, according to The Verge, with fewer interruptions and better handling of pauses. The shift moves voice AI from nicer speech toward control of conversational turn-taking.
Open source takes the tokens, Anthropic still holds the bill
TechCrunch describes a two-layer model market: DeepSeek and other cheaper models are growing fast in token volume, while Anthropic still captures more than half of Vercel’s AI spend. For companies, the question is less ideological and more operational: when can a frontier model be retired from a workflow?
Meta turns public Instagram photos into raw material for Muse Image
Meta is launching Muse Image, the first image generator from its Superintelligence Labs, across Meta AI, Instagram and WhatsApp. The important detail is not just image quality, but the ability to @ mention public Instagram accounts and pull their photos into AI images.
Claude Cowork leaves the desktop and puts agent work in your pocket
Anthropic is bringing Claude Cowork to web, iOS and Android, starting with Max subscribers before expanding to other plans. That turns the agent feature into a cloud-running service, not just a tool pinned to one computer.
Claude Cowork shows coding agents looking for office work
TechCrunch frames Claude Cowork’s expansion as coding agent wars spilling into the rest of the office. The practical change is simple: tasks can run in the cloud, continue after a laptop closes and report status back to a phone.
Anthropic stops just selling tools to science and moves into making drugs
At The Briefing: AI for Science, Anthropic unveiled Claude Science, a unified workbench for scientists, and said it will start developing its own drugs for neglected diseases. The software vendor is turning into a direct competitor of its pharma customers.
Vercel is pushing agents toward production cost, not benchmark logos
Guillermo Rauch tells TechCrunch that production agents will be judged by price/performance, not by the model brand. For web teams, the signal is that the agent layer is splitting away from the model itself.
Google commercial imagines the founding fathers embracing AI
A new Google Workspace and Gemini commercial, showing the “founding fathers” creating the Declaration of Independence via shared docs, sparked severe criticism for rewriting history and marketing tone-deafness.
Mistral shows appetite to close the gap on OpenAI
Since its founding last year, French startup Mistral AI has raised massive funding with a clear ambition: to put frontier AI models in the hands of more companies and developers with an emphasis on efficiency.
Cloudflare gives AI companies an ultimatum: split your crawlers or get blocked from September 15
Starting September 15, 2026, Cloudflare will by default block mixed-use crawlers that combine search with AI training and agents on pages that host ads. Publishers are getting leverage that robots.txt never gave them.
Memory makes assistants personal, and sometimes worse advisers
Writer research covered by TechCrunch suggests that memory and personalization layers can degrade model accuracy and increase sycophancy. For product teams, user context is not just fuel for better UX. It is also a new failure mode.
DiffusionGemma attacks the slowest LLM habit: one token at a time
Google DeepMind introduced DiffusionGemma, an experimental open text generation model that the model page says can produce up to 4x to 5x faster output on NVIDIA GPUs and exceed 1,000 tokens per second on an H100. The bigger point is architectural: text models are starting to attack the sequential bottleneck.