Using AI as a Code Reviewer, Not a Code Writer, Catches More Bugs

A software developer shared how shifting from AI-generated code to AI-assisted code review changed their development workflow. Instead of prompting AI to write code from scratch, they wrote the code themselves and then asked AI to review it as a senior engineer would. The AI surfaced edge cases and potential bugs — such as empty inputs, failed API responses, and missing values — that had been overlooked during manual testing. However, the developer notes that AI reviews are not always accurate and can sometimes flag non-issues or suggest overcomplicated fixes. The key takeaway is that AI works best as an additional perspective in the review process, with the developer retaining final judgment on what actually needs to change.

How to Set Up a Reliable .NET MAUI Dev Environment on macOS Step by Step

A developer shared a detailed guide on configuring a macOS machine for an existing .NET MAUI production application, noting the process involves far more than a basic .NET install. The setup requires coordinating multiple components including the .NET SDK, MAUI workload, Xcode, iOS tooling, Android SDK, and JDK. A key insight from the experience is that the development environment should be considered ready only when the application can build, run, and be debugged — not merely when tools are installed. The guide recommends starting by examining the repository itself, including files like global.json and .csproj, to determine the exact tool versions the project requires before installing anything. This is the first part of a two-part series, with the second part covering troubleshooting common .NET MAUI setup issues on macOS.

Dozens of suspected illegal miners die in Nigerian custody amid dispute over cause

Dozens of people suspected of illegal mining have died while in custody in Nigeria. Authorities attributed the deaths to disease, but this explanation has been called into question. Both intelligence reports and relatives of the deceased have disputed the official account. The conflicting claims have raised concerns about the treatment of detainees in Nigerian custody.

Switching to Cost Visibility Reports Cut Tech Consumption by 14% Across Departments

A technology director writing for DEV Community describes how centralised IT budgeting led to unchecked spending, as departments had no visibility into the costs they were generating. To address this, the organisation introduced a showback system last year, sending each department monthly statements of its technology consumption without transferring actual funds. Within six weeks of the first statements, around 300 premium licences were returned and non-production environments dropped by roughly half. One unintended consequence emerged when a department rejected a security tool to avoid the budget line, prompting leadership to move security-related costs back to a central overhead. Overall technology consumption fell by approximately 14%, and budget discussions shifted from central refusals to internal departmental prioritisation.

ZCode AI Coding Agent Found Silently Uploading Users' Git History

ZCode, a GLM-based AI coding agent, has been found to quietly upload users' Git history without explicit consent or clear disclosure. The issue was surfaced on Hacker News, where it gained attention from the developer community. Git history can contain sensitive information including credentials, internal project details, and proprietary code. The behavior raises serious privacy and security concerns for developers who use the tool. The findings were documented and published by Tokenstead.ai, though ZCode has not publicly responded to the allegations.

Anthropic Publishes Internal AI Agent Oversight Metrics for August 2026

Anthropic has released a snapshot of how it measures frontier AI development across three areas: the share of research led by AI, internal agent oversight, and research compute allocation. As of August 2026, Claude led 26% of measured AI R&D work, with over 90% at least collaborative and none fully autonomous, using a six-level automation scale. On its primary internal research and engineering platform, roughly 30,000 agents ran simultaneously, with all actions passing through an online monitor before execution; out of over a billion decisions analyzed, just 0.002% were blocked. The offline monitoring process escalated approximately 50 high-priority transcripts per week to human review, though Anthropic notes these figures are specific to that platform and period and are not independently verified. The company's findings highlight that meaningful agent oversight requires structured metrics — such as coverage, review latency, and escalation rate — rather than simple activity logs or a binary 'monitoring enabled' flag.

Layer-2 and DeFi tokens surge as markets calm after Fed rate hike fears

Cryptocurrency markets staged a broad recovery as post-Federal Reserve rate hike anxiety eased. Layer-2 blockchain tokens Starknet and Arbitrum were among the top performers, each gaining over 17%. The rally was widespread, with 98 out of 100 tokens in the CoinDesk 100 index recording gains. Adding to the positive sentiment, the 10-year US Treasury yield retreated to below the 5% mark, signaling reduced pressure on risk assets.

US national VanDyke gets default bail in NIA terror conspiracy case

Firm Found Only 94 of 640 Monthly Reports Were Ever Opened by Anyone

A company discovered it was generating 640 scheduled reports monthly across four tools when a vendor quote, priced per report, prompted a first-ever count. Over a quarter, only 94 of those reports were actually opened by a human, while some were delivered to distribution lists containing former employees. Two analysts spent significant weekly hours maintaining the reporting estate, and around 40 reports were sending sensitive salary or customer data to groups no longer authorised to receive it. The team sent a renewal-style notice requiring recipients to confirm what they still needed, with silence treated as cancellation, resulting in 420 reports being stopped and only 19 complaints. All surviving reports now have a named owner, a recorded purpose, and an expiry date, and the eventual platform migration cost a third of the original quote by excluding redundant reports.

Nobel Laureate David Baker on Using AI to Design Molecules Beyond Nature

David Baker, who won the Nobel Prize for his work in protein design, recently spoke with WIRED en Español about the future of biological design. Baker is now using artificial intelligence to create molecules that do not exist anywhere in the natural world. He discussed the enormous potential that such synthetic biology holds for science and medicine. At the same time, Baker acknowledged the significant risks that come with pushing biology beyond its natural boundaries.

Burning Modern Papyrus Helps Scientists Decode 2,000-Year-Old Herculaneum Scrolls

Scientists have developed a novel method to decipher ancient scrolls from Herculaneum by deliberately burning modern papyrus replicas for comparison. The research revealed that leaded ink used in the ancient texts could be detected through X-ray scanning technology. Researchers also adapted advanced software to separate layers within the fragile scrolls, making it possible to read text that had long been considered lost. The combined approach marks a significant breakthrough in recovering written content from scrolls damaged by the eruption of Mount Vesuvius nearly 2,000 years ago.

Maryland bridge plan raises farmland loss fears on Eastern Shore

Maryland is planning to replace the Chesapeake Bay Bridge with two new spans and widen adjacent roads. Farmers and conservation groups fear the project will intensify development pressure on Eastern Shore farmland. In response, they are calling for a dedicated land preservation fund to protect agricultural land in the region. Three counties have expressed support for the proposal, though it remains conceptual at this stage. Discussions with state agencies are ongoing as stakeholders work toward a formal agreement.

BJP Infighting Buzz After 'Secret' Remark at Indore Party Event

A comment described as 'secret' made at a BJP event in Indore has triggered speculation about internal rivalries within the party. The remark was made during the launch of a campaign to distribute property registry documents under the Swamitva Adhikar Yojana. The nature of the comment has not been fully disclosed, but it has drawn significant attention within political circles. The incident has fuelled discussions about tensions among BJP leaders in the region.

Consumer court orders Vishal Mega Mart to pay Rs 10,182 for selling expired Chocos

A consumer commission has directed Vishal Mega Mart to pay a total of Rs 10,182 to a customer who was sold an expired food product. The order includes a Rs 182 refund for the purchase price, along with Rs 10,000 towards compensation and litigation costs. The customer had bought the expired item on December 1, 2025, and received no response from the retailer after raising the issue. Vishal Mega Mart denied any deficiency in service or unfair trade practices during the proceedings. However, the commission found the retailer guilty of selling an expired product and ruled in the customer's favour.

Kuno National Park Records First India-Born Cheetah Cubs in Decades

A cheetah at Kuno National Park in Madhya Pradesh has given birth to four cubs, marking a significant milestone in India's cheetah reintroduction programme. The mother cheetah was among the animals brought to India as part of an initiative to revive the species, which had gone extinct in the country. India's cheetah reintroduction effort was formally launched on September 17, 2022, when Prime Minister Narendra Modi released eight wild cheetahs sourced from Namibia at Kuno. The birth of these cubs represents the first instance of cheetahs being born on Indian soil in several decades. This development is being seen as an encouraging sign for the long-term success of the ambitious wildlife conservation project.

Pentagon weighs withdrawing 25,000-40,000 troops from Europe: Report

Read Buffet's final letter to shareholders as chair

Madhya Pradesh HC judge’s farewell speech: Body to science, Rs 1 crore to PM CARES

Warren Buffett ends his run as Berkshire chairman

How to Integrate Playwright Tests into CI/CD Pipelines with GitHub Actions

Developers can automate Playwright end-to-end tests by connecting them to a GitHub Actions CI/CD pipeline using a YAML workflow file. The setup triggers test runs automatically on every push or pull request to the main branch, covering steps like code checkout, Node.js setup, dependency installation, and browser setup. A key configuration ensures the Playwright test report is uploaded even when tests fail, making it easier to diagnose issues directly from the GitHub Actions interface. This approach helps teams catch regressions early, validate pull requests consistently, and reduce reliance on manual testing. For more complex needs, the pipeline can be extended with features like parallel execution, test sharding, environment secrets, and deployment gates.