doorss

Google is massively increasing its financial backing of Anthropic, with plans to invest up to $40 billion to help the AI lab expand its computing infrastructure.

Anthropic confirms that widespread complaints about degraded Claude Code quality over the past two months were grounded in real problems, and details the issues in a postmortem.

OpenAI released GPT-5.5, unifying Codex and the main model into a single system with gains in agentic coding, computer use, and general task performance.

OpenAI explains why it has stopped evaluating on SWE-bench Verified, arguing the benchmark no longer meaningfully differentiates frontier coding models.

DeepSeek released V4 preview models, and LMSYS details day-zero benchmarks and infrastructure work to support fast inference and verified reinforcement learning.

Axios reports that AI agent costs are in some cases exceeding the cost of human workers, challenging assumptions about AI-driven cost savings.

The FT argues that whether AI can perform a task is only a small part of the employment picture, and the narrative around mass AI job displacement overlooks important factors.

Anthropic's survey of 81,000 Claude users finds that workers in AI-exposed roles have greater concerns about job displacement, with concerns also higher among lower-income respondents.

Anthropic's Economic Research team is launching a monthly survey conducted through Anthropic Interviewer to better understand AI's real-world economic impact beyond quantitative data.

The article draws parallels between the West's deindustrialization and a similar loss of coding skills, arguing that over-reliance on AI tools risks hollowing out software engineering capability.

Researchers discovered a stable IndexedDB-based identifier in Firefox that could link all of a user's private Tor browsing identities together.

OpenAI introduced a privacy filter feature designed to give users more control over how their data is handled during interactions with its models.

A report reveals that Claude Desktop silently installs browser extensions in Chrome and other browsers without informing users, raising privacy concerns.

The article argues that the EU's age verification initiative is a vehicle for introducing mandatory digital identity infrastructure with broad surveillance implications.

Bluesky published technical details on how it serves its "For You" feed, highlighting the platform's unique architecture where anyone can run their own custom feed implementation.

Bluesky experienced a sustained DDoS attack lasting nearly a full day, causing service disruptions on the platform.

TV advertising is gaining renewed value as streaming platforms increasingly adopt ad-supported tiers to generate new revenue streams.

A leaked deck from StackAdapt reveals how OpenAI's ad partner is selling ad placements inside ChatGPT based on prompt relevance, signaling AI-native advertising.

The FT explores how AI is reshaping marketing roles and how live sports events are becoming increasingly valuable for advertisers seeking engaged audiences.

S&P Global Ratings muutti Suomen luottoluokituksen näkymät negatiivisiksi. Syynä ovat julkiseen talouteen kohdistuvat riskit.

Lidlin omistajan Dieter Schwarzin firma sijoitti suomalaiseen Confidental Mindiin osana pyrkimystä vahvistaa Euroopan digitaalista riippumattomuutta.

Pieni suomalainen kunta onnistui tekoälykokeilussa, mutta Sitran mukaan Suomi uhkaa jäädä jälkeen, koska yksittäisiä onnistumisia ei osata skaalata koko julkishallintoon.