doorss

Monday, May 18, 2026 — 18 items

AI & LLMs: Research, Capabilities, and Industry

Anthropic Research · www.anthropic.com
Anthropic released a paper outlining its views on US-China AI competition, arguing it is essential for the US and allies to maintain a lead over authoritarian governments like the CCP in AI development.
Hacker News · www.anthropic.com · comments
Anthropic published research on natural language autoencoders, a technique for converting Claude's internal representations into human-readable text to improve interpretability.
Financial Times · www.ft.com
Anthropic will discuss the capabilities of its Mythos AI model with Financial Stability Board members, after the model exposed significant cyber vulnerabilities in financial infrastructure.
Lobsters · www.seangoedecke.com · comments
DeepSeek's V4-Flash model has revived interest in LLM steering vectors as a practical technique, because its architecture makes activation-level interventions more effective.
Lobsters · sites.google.com · comments
Research examining how LLM-generated text is systematically distorting patterns in written language, documenting measurable shifts in vocabulary and style across online content.

AI's Impact on Software Engineering and Work

Hacker News · frederickvanbrabant.com · comments
A post arguing that AI tools speed up coding but don't necessarily accelerate overall software delivery processes, because bottlenecks often lie elsewhere.
Hacker News · twitter.com · comments
Mitchell Hashimoto observes that some companies are in a state of "AI psychosis," making irrational decisions driven by hype rather than sound engineering judgment about AI capabilities.
Financial Times · www.ft.com
Amazon employees are using the company's in-house MeshClaw AI tool for unnecessary tasks to climb internal AI usage leaderboards, highlighting perverse incentives in corporate AI adoption.

Privacy, Security, and Surveillance

Hacker News · blog.mozilla.org · comments
Mozilla submitted a formal response to UK regulators arguing that VPNs are fundamental privacy and security tools and should not be undermined by proposed regulations.
Lobsters · www.techspot.com · comments
A security researcher claims Microsoft built a secret backdoor into BitLocker encryption and has released a tool called YellowKey that demonstrates the vulnerability.
Hacker News · macdailynews.com · comments
The US Department of Justice is demanding Apple and Google reveal the identities of over 100,000 users of a car modification app as part of an emissions enforcement crackdown.

Bluesky and the 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), enabling decentralized music tracking.
Simon Willison · atproto.com
The AT Protocol team published technical details on how Bluesky's algorithmic "For You" feed works, explaining how anyone can implement and run their own custom feed.

Suomi / Finland

Yle pääuutiset · yle.fi
Puolustusvoimat varautui torjumaan Ukrainan iskuun Venäjälle liittyneet droonit, koska ne olisivat Uudellamaalla voineet aiheuttaa isoja vahinkoja.
Yle pääuutiset · yle.fi
Suomi oli läpi kevään ennakkosuosikki Euroviisuissa, mutta jäi lopulta kuudenneksi. Bulgaria voitti kilpailun.
Yle pääuutiset · yle.fi
Metan tekoälylaseja on käytetty naisten salakuvaamiseen, ja myös yhtiön työntekijät näkevät arkaluontoista materiaalia. Huolet älylasien väärinkäytöstä lisääntyvät.

Advertising Technology and Video Generation

Financial Times · www.ft.com
ByteDance and Kuaishou are outpacing Western competitors in AI video generation quality, with the technology lifting video quality across advertising and entertainment sectors.
Hacker News · www.buchodi.com · comments
A detailed breakdown of ChatGPT's advertising attribution loop, revealing how OpenAI's ad partner StackAdapt is selling placements based on prompt relevance to advertisers.
My preferences