doorss

Anthropic published a policy paper on building trustworthy AI agents, addressing how the shift from chatbots to autonomous agents (like Claude Code) raises new challenges around reliability and safety in real-world deployments.

US Treasury Secretary Scott Bessent convened top bank CEOs to discuss cybersecurity risks posed by Anthropic's latest AI model, which has reportedly detected decades-old vulnerabilities in financial systems.

Anthropic is rapidly closing the gap with OpenAI in enterprise adoption, driven largely by strong interest in its Claude Code product among US businesses.

Executives in finance and cyber defence are mapping where Anthropic's Claude plug-ins will and won't replace human judgment, betting that trust remains a key differentiator in white-collar professions.

Simon Willison highlights that OpenAI's ChatGPT voice mode runs on a much older, weaker model with an April 2024 knowledge cutoff, which is non-obvious to most users who assume the voice AI is the smartest version.

Law firms may raise prices for fixed-fee contracts as clients use AI to generate large volumes of queries and correspondence, increasing the workload on lawyers.

The Linux kernel project has published official documentation on the use of AI coding assistants when contributing to the kernel, establishing guidelines for how developers should handle AI-generated code.

A project on GitHub demonstrates reverse engineering of Google's SynthID watermarking system used to detect AI-generated content from Gemini, raising questions about the robustness of steganographic AI watermarks.

The Electronic Frontier Foundation announced it is leaving X (formerly Twitter), a notable move by one of the most prominent digital rights organizations. This follows a broader trend of organizations migrating away from the platform.

Big Brother Watch reports that Apple's latest iPhone update implements content restrictions in the UK that limit internet freedom, raising privacy and censorship concerns.

The CPUID website distributing popular system utilities CPU-Z and HWMonitor was hijacked, potentially serving compromised downloads to users of these widely-used diagnostic tools.

Tekoäly toistaa sukupuolistereotypioita esimerkiksi palkkaehdotuksissa. Vinoumat voivat vaikuttaa myös sairauksien diagnosointiin.

Artikkeli kertoo kahdesta suomalaisesta, joilla on hyvin erilainen suhde tekoälyyn: toiselle se on ainoa ystävä, toinen käyttää sitä ihmissuhteiden parantamiseen.

Helsingin yliopiston tutkimus osoitti ensimmäistä kertaa yhteyden varhaisen ADHD-diagnoosin ja myöhemmän paremman koulutustason välillä.

A suspect was arrested after throwing a Molotov cocktail at Sam Altman's home and threatening to burn down OpenAI's San Francisco headquarters.

Sam Altman published a blog post responding to the attack on his home, which drew significant discussion on Hacker News.

Stripe describes their approach to selective test execution that enables fast continuous integration across their massive 50-million-line Ruby monorepo.

Colin Percival reflects on 20 years of building on AWS, discussing the culture of ownership and broad responsibility that has defined his experience on the platform.