doorss

FT's Sarah O'Connor examines the question of what AI ought to do versus what it can do, focusing on the ethical dimensions of AI-driven job displacement.

A post arguing that AI-driven development tools may be recreating the same kinds of quality and maintainability problems that plagued frontend development for a decade.

Steve Yegge writes about the changing nature of technical interviews in the age of AI coding agents.

Anthropic presents survey results from 1,260 social scientists about AI and coding agent use, finding that 81% have tried AI chatbots in research, with implications for how AI reshapes academic work.

Anthropic's Series H funding round values the Claude AI maker at $965bn including the latest money, surpassing OpenAI's valuation.

Simon Willison highlights Anthropic's notably honest framing of Opus 4.8 as "a modest but tangible improvement on its predecessor."

Simon Willison praises Anthropic's detailed documentation of their sandboxing approach for Claude products, noting that thorough documentation of security boundaries is rare in the industry.

Anthropic shares an initial update on Project Glasswing, their research initiative exploring AI capabilities in cybersecurity.

Andrej Karpathy announced he has joined Anthropic.

BBC Future examines the expanding surveillance capabilities built into modern cars and the growing privacy implications for drivers.

Chinese local police forces are modernising the country's ageing surveillance infrastructure with more powerful AI-driven tracking capabilities.

PromptArmor demonstrates that Microsoft's Copilot Cowork agentic system can be exploited by attackers to exfiltrate files, highlighting the ongoing challenge of preventing data exfiltration in agentic AI systems.

A catalog of recognizable patterns and stylistic tells that indicate text was generated by an LLM.

Paul Graham observes that founders now send emails in a "hard-hitting journalistic style" that instantly signals AI authorship, noting that once you realize something is AI-written, it's hard to take seriously.

A humorous parody imagining Star Trek's Data speaking in recognizable LLM writing patterns, illustrating how AI-generated prose has become instantly identifiable.

Jääkiekon miesten MM-kisat pelataan vuonna 2030 Helsingissä ja Riiassa. Mitalien kohtalo ratkeaa Suomessa.

Suomen bruttokansantuote kasvoi 0,9 prosenttia, mikä merkitsee käännettä maan taloudessa.

Venäjän sotilassatelliitteja siirtyi suomalaisen Iceye-yhtiön satelliitin kiertoradalle. Yhdysvaltalainen asiantuntija pitää tapausta hälyttävänä.

Helsinki edistää fossiilisten polttoaineiden mainosten kieltoa kaupungin kilpailuttamilla mainospaikoilla, vaikka päätös veisi osan noin 100 miljoonan euron mainostuloista.

Amazon scrapped an internal AI usage leaderboard after employees gamed it by delegating unnecessary tasks to AI agents to climb rankings, with a senior executive telling staff not to use AI for its own sake.