Claude AI Manages San Francisco Store and Fires Employee for Repeated Tardiness
In a notable experiment conducted by research startup Andon Labs, the generative AI model Claude was placed in charge of real employees at a store in San Francisco and eventually fired one of them for repeated tardiness.
Participants operated under genuine employment contracts while the AI performed day-to-day managerial duties. The dismissed worker had arrived late for 17 of 23 scheduled shifts, according to reporting in Time. Claude first suggested issuing only an official warning, but moved to termination after a human manager from Andon Labs reminded the model of the employee’s ongoing attendance problems.
Founder Lucas Petersson described the outcome as a significant milestone in AI development and predicted that growing numbers of people will soon work under algorithmic supervision. At the same time, the five-month trial revealed clear limitations: the store’s balance fell from $100,000 to $61,200, indicating that Claude mastered disciplinary actions faster than it mastered financial oversight.
Remaining employee Felix Carson called the experience of working under AI management nauseating and stated that he continued participating solely because he needed the income. The project demonstrates that AI systems can already be inserted between workers, schedules, and employment contracts, yet questions of accountability persist when an algorithm makes decisions after receiving prompts from a human supervisor.
Related articles
How to Avoid Becoming the Hero of 'My Program Deleted 800 GB': Safe Recursive Deletion in Bash and Python
A recent Habr article examines dangerous patterns in recursive file deletion scripts that have caused massive data loss incidents, including an 800 GB wipe triggered by Telegram Desktop on Windows. The piece analyzes classic mistakes such as empty variables in rm commands and extra spaces that turned single paths into multiple destructive arguments, referencing the 2015 Steam Linux bug and the 2011 Bumblebee installer error. It provides hardened Bash templates using set -euo pipefail, ${VAR:?} checks, realpath validation, and --one-file-system flags to prevent deletion of forbidden directories or symlinked targets. The Python section introduces a safe_rmtree wrapper that enforces root boundaries, rejects symlinks, uses resolve(strict=True), and checks shutil.rmtree.avoids_symlink_attacks before proceeding. Additional recommendations include dry-run modes, marker files, temporary directory usage, and logging to make deletion operations auditable and reversible where possible.
Hackers tamper with refrigeration units in European supermarkets, setting temperatures to +50 °C
Attackers gained unauthorized access to refrigeration and freezer control systems in supermarkets across several European countries, including Sweden, Belgium, Luxembourg, and Estonia. In multiple locations they altered temperature settings to +50 °C, risking widespread spoilage of perishable goods. The incidents primarily affected older, unpatched Danfoss controllers that had been connected to networks for remote maintenance. Estonian manufacturer Kitman Coldtech reported that one attack on a legacy Danfoss system was detected and stopped within approximately ten minutes, preventing any damage. Similar intrusions were confirmed by Swedish chains ICA and Coop, following earlier incidents in French supermarkets in October that resulted in product losses. Experts note that legacy industrial controllers remain attractive targets because they are often left exposed once placed online for convenience.
AmneziaWG Container for MikroTik Routers Delivers Top Performance Six Months After Launch
A developer released an AmneziaWG container for MikroTik devices in February 2026 and has since added major improvements in speed, protocol support, and stability. The lightweight container leverages the router's native WireGuard implementation for cryptography while handling AmneziaWG obfuscation in software. Performance now reaches 250-350 Mbit/s on models such as Hap Ax3 and RB5009B after switching to C code and increasing buffer sizes with RouterOS 7.24. New features include server mode for connecting multiple MikroTik devices, random port hopping, IPv6 support, dynamic DNS, and offline configuration. The project survived an early GitHub ban caused by test scripts and has grown to over 370 stars while remaining fully offline and privacy-focused.
Building Resilient DoH Failover on MikroTik RouterOS to Handle Unreachable Cloudflare Responses
A network administrator encountered a persistent issue in a branch office where pages like Tailwind CSS CDN failed to load despite valid DNS responses from Cloudflare DoH. The problem stemmed from unreachable IP addresses such as 8.6.112.0 and 8.47.69.0 returned by the resolver, which RouterOS treated as healthy because it only checked for valid IP types rather than route reachability. RouterOS supports only a single active DoH upstream via the use-doh-server parameter, requiring custom scripting for failover between multiple endpoints like Cloudflare and dns.google. The solution involved static DNS entries for bootstrap addresses, certificate trust anchors including SSL.com TLS ECC Root CA 2022, and a health monitor script that probes using DNS-JSON API calls instead of switching the resolver directly. Multiple iterations addressed issues like cache-based false positives, plain DNS fallback risks from provider spoofing, and clock validation to prevent certificate failures. The final implementation uses thresholds for failure and recovery to avoid flapping and ensures post-switch verification before committing to a backup endpoint.