Blog
Topic: AI
17 posts on this topic.
Cruelty Toward Claude: What Anthropic's New Usage Rule Covers
From November 12, Anthropic prohibits sustained, needless cruelty toward its models. What the rule covers, what it leaves out, and why it is still contested.
Product Owners and AI Agents: Who Writes the Requirements Now?
Coding agents build what the requirement says and guess the rest. What that shifts for product owners, and what stays with people.
When Your Agent Represents You Online
Browser agents already act in your name - but with your session instead of their own identity. Why real representation needs mandates, labeling and clear law.
When Your AI Agent Pays: Money on Two Rails
Visa, Mastercard and Stripe are building payment infrastructure for AI agents while Coinbase bets on stablecoins with x402. How agents will pay in our name.
Human or Machine: Who Is Acting There Online?
AI agents act autonomously on the web. Humans now need personhood credentials, agents need mandates - and both require an identity of your own.
The EU AI Act in Practice: What Applies to App Developers in 2026/2027
Chatbot disclosure from August 2026, high-risk rules from late 2027: the AI Act timeline after the Digital Omnibus and a to-do list for app developers.
AI-Assisted Programming One Year Later: What's Left of the Promises
19 percent slower instead of faster, declining trust, more duplicated code: the AI coding balance sheet per METR, Stack Overflow and GitClear.
The agent won't object: cleaning up instructions in grown projects
Coding agents resolve contradictory rules silently, without warning you. A tutorial on cleaning up the instruction files of grown projects.
Local LLMs on the Mac: What Apple Silicon Realistically Delivers
Language models without the cloud: why Apple Silicon is suited for it, which model sizes run on 16, 32 and 64 GB of RAM, and where the limits are.
tokensave: An MCP Server That Saved Me Millions of Tokens While Coding
tokensave is an MCP server that gives Claude Code a local code graph instead of file reads. What it does, the 94 percent benchmark, and where it falls short.
What Is Private Cloud Compute?
What Private Cloud Compute is, how its architecture works, what Apple's security whitepaper proves and what the move to Google servers changes.
Project Panama: Why an AI Company Cut Apart Millions of Books
Anthropic's Project Panama, uncovered by the Washington Post: millions of books cut apart and scanned for AI training - court records and the settlement.
The Power Grid Under Attack: AI as Attacker and Defender
Millions of connected inverters and batteries make the power grid vulnerable. How AI-driven attacks could trigger blackouts - and why defense needs AI too.
Between Disruption and Demographics: AI and the German Labor Market
Is AI killing jobs in Germany? The data points elsewhere: jobs are changing, not disappearing - and demographics weigh heavier than automation.
AI App Flood: Almost Nobody Wins
Since AI coding tools became routine, iOS app releases have exploded - but usage hasn't. Why shipping alone stopped being a competitive advantage.
The Illusion of Digital Relief: Why the Skills Shortage Persists Despite AI
AI and digitalization don't automatically solve the skills shortage. Why efficiency gains often mean more work, not less - an analysis.
Vibe Coding: Democratizing Software or the End of Quality?
What vibe coding delivers and where it breaks: Veracode and METR data, the Replit incident - and why developers become architects and auditors.