doorss

Anthropic published a research article on building trustworthy AI agents, addressing the shift from simple chatbots to autonomous agents like Claude Code and how to ensure their reliability and safety in practice.

Simon Willison shares Steve Yegge's account of a conversation with a Google tech director about the company's internal AI adoption, describing it as the 'craziest convo' of the year.

Bryan Cantrill argues that LLMs inherently lack the virtue of laziness — they don't optimize for future time and will produce unnecessary work, which poses a problem for software development.

Cantrill's full blog post explores how LLM-generated code threatens software quality because AI lacks the programmer's instinct to minimize unnecessary complexity and future maintenance burden.

Tekoäly toistaa sukupuolistereotypioita esimerkiksi palkkaehdotuksissa. Vinoumat voivat vaikuttaa vakavasti myös esimerkiksi sairauksien diagnosoinnissa.

A project that reverse-engineers Google's SynthID watermarking system used to detect AI-generated content from Gemini models, touching on steganographic techniques for embedding hidden signals in AI outputs.

Android has introduced a feature that strips location metadata from photos by default when sharing, preventing inadvertent disclosure of where photos were taken.

An attacker purchased 30 WordPress plugins and injected backdoors into all of them, highlighting serious supply chain security risks in the WordPress ecosystem.

US Treasury Secretary Bessent convened bank CEOs to discuss cybersecurity risks after Anthropic's latest AI system detected decades-old vulnerabilities in critical financial infrastructure.

Merivartiosto kertoo, että GPS-häirintä Suomenlahdella on jatkunut voimakkaana. Öljytankkereiden liikenne on palannut drooni-iskuja edeltävälle tasolle.

A visual tour of Helsinki's Oodi central library, showcasing the building's architecture and public services.

Suomi kehittää terveydenhuoltoa, jossa tekoäly ja data tunnistavat potilaiden sairausriskit ennakkoon. Tavoite vaatii myös lakimuutoksia.

OpenAI investors are questioning the company's $852bn valuation as CEO Sam Altman refocuses strategy, while Anthropic tests its early lead in business adoption.

Anthropic is rapidly closing the gap with OpenAI in US business adoption, driven by strong demand for its Claude Code product.

Executives across finance and cybersecurity are mapping where Anthropic's Claude plug-ins will and won't replace human judgment, with trust emerging as the key differentiator for white-collar work.

An argument that configuration flags accumulate over time and become a source of software rot, making codebases harder to maintain and reason about.

A developer describes finding a bug in a program that had been formally verified as correct using the Lean theorem prover, raising questions about the gap between formal proofs and real-world behavior.

The Electronic Frontier Foundation announced it is leaving the X platform, a notable move by a major digital rights organization that may signal continued migration to alternatives like Bluesky.

Google Search is introducing a new spam policy targeting websites that hijack the browser's back button, a widely despised dark pattern used by many ad-heavy sites.