doorss

Sunday, May 17, 2026 — 18 items

AI Industry & Research

Anthropic Research · www.anthropic.com
Anthropic released a paper on US-China AI competition, arguing it's essential that the US and allies maintain a lead over authoritarian governments like the CCP in AI development.
Anthropic Research · www.anthropic.com
The Anthropic Institute shares its research agenda, using information from within a frontier lab to investigate AI's impact on the world and sharing findings publicly.
Hacker News · www.anthropic.com · comments
Anthropic published research on natural language autoencoders that convert Claude's internal representations into human-readable text, advancing interpretability of LLM reasoning.
Hacker News · www.anthropic.com · comments
Anthropic research on training Claude to understand the reasoning behind its instructions, aiming to improve the model's ability to generalize from guidance rather than just follow rules.
Financial Times · www.ft.com
ByteDance and Kuaishou are outshining western rivals in AI-generated video quality, with applications spanning advertising and entertainment.

AI's Impact on Software Engineering & Work

Hacker News · twitter.com · comments
Mitchell Hashimoto argues that some companies are in a state of "AI psychosis," making irrational decisions about AI adoption disconnected from practical reality.
Simon Willison · simonwillison.net
Simon Willison reflects on how AI coding tools are making programming language lock-in far less significant, as demonstrated by Bun's migration from Zig to Rust.
Financial Times · www.ft.com
Bug bounty programs are being overwhelmed by a surge of spurious AI-generated vulnerability submissions, straining the resources of security teams.
Financial Times · www.ft.com
Amazon employees are using the company's internal AI agent tool MeshClaw for unnecessary tasks to climb AI adoption leaderboards, gaming the metrics intended to track genuine AI adoption.

Privacy & Security

Hacker News · macdailynews.com · comments
The US Department of Justice is demanding Apple and Google identify over 100,000 users of a car modification app as part of an emissions enforcement crackdown.
Hacker News · tmctmt.com · comments
Research shows that Mullvad VPN exit IP addresses can serve as a fingerprinting vector, potentially compromising user anonymity despite the VPN's privacy-focused design.
Lobsters · www.abgeo.dev · comments
Security researcher discovered that a cheap smart doorbell has vulnerabilities allowing fleet-wide account takeover and call hijacking across all devices.

Bluesky & AT Protocol

Hacker News · tangled.org · comments
Rocksky is a new music scrobbling and discovery service built on the AT Protocol (Bluesky's underlying infrastructure), bringing Last.fm-style functionality to the decentralized social web.

Suomi / Finland

Yle pääuutiset · yle.fi
Puolustusvoimat varautui torjumaan Ukrainan iskuun liittyneet droonit, jotka olivat matkalla Suomea kohti Venäjän suunnasta ja olisivat voineet aiheuttaa isoja vahinkoja Uudellamaalla.
Yle pääuutiset · yle.fi
Suomalaistaustainen professori kehottaa Eurooppaa luomaan omat digijättinsä. Oura-älysormuksen valmistaja on siirtänyt pääkonttorinsa Yhdysvaltoihin.
Yle pääuutiset · yle.fi
Metan tekoälylasien väärinkäytöstä on kasvavia huolia, sillä laseilla on kuvattu naisia salaa ja arkaluontoista materiaalia päätyy myös työntekijöiden nähtäville.

Advertising Technology & Personalised Pricing

Hacker News · www.buchodi.com · comments
Analysis of how ChatGPT's ad serving works, detailing the full attribution loop from prompt to ad placement and tracking.
Hacker News · www.adweek.com · comments
A leaked deck reveals that StackAdapt, OpenAI's ad partner, is selling ChatGPT ad placements targeting users based on the relevance of their prompts.
My preferences