‘Bribe to suppress rape case’ allegation: Mann’s wife sends notice to ex-judge

Deepika Padukone exits Spirit film amid profit-share dispute, Triptii Dimri steps in

Bollywood actress Deepika Padukone has exited director Sandeep Reddy Vanga's upcoming film 'Spirit'. Reports indicate her departure was linked to a demand for a ten percent share in the film's profits. The development has sparked debate around gender bias in the Hindi film industry. Fans have expressed support for Padukone on social media platforms. Actress Triptii Dimri has since been brought on board to replace her in the project.

Sangakkara restores 120-year-old villa inside Sri Lanka's Galle Fort

Cricket legend Kumar Sangakkara and his wife have restored a historic villa located within Galle Fort, Sri Lanka. The property, which is approximately 120 years old, has been carefully renovated to preserve its weathered walls and layered patina while incorporating modern elements. The residence now features four bedrooms, courtyard gardens, and a rooftop pool. The restored villa offers guests an immersive experience of Galle Fort's rich architectural and cultural heritage.

Iran allows Iraqi oil tankers through Hormuz

Jaishankar to Visit Russia on Aug 23-24 for Trade and Scientific Talks

Indian External Affairs Minister S. Jaishankar is scheduled to visit Russia on August 23 and 24. The visit is expected to focus on bilateral trade and scientific cooperation between the two countries. The trip signals continued diplomatic engagement between India and Russia. Further details about the agenda or meetings have not been provided in the available information.

Saab Unveils Stealthy Supersonic Combat Drone Designed to Fly Alongside Gripen E

Swedish defense company Saab has revealed a concept for a stealthy, supersonic unmanned combat aircraft intended to operate alongside its Gripen E fighter jets. The autonomous drone is designed to carry out a range of combat missions while maintaining a low radar signature. Two technology prototypes, designated A1 and A2, are currently undergoing testing to advance the aircraft's flight systems. The program aims to establish an effective human-machine teaming model to enhance overall battlefield performance.

WordPress Plugin CVE-2026-13736 Exposes Member PII to Unauthenticated Users

A medium-severity vulnerability, CVE-2026-13736, has been identified in the NewPath WildApricotPress Member Directory WordPress plugin up to and including version 1.0.0. Discovered by cybersecurity researcher Huynh Kien Minh, the flaw stems from a publicly accessible REST API endpoint that lacks proper access controls. Because the backend serializes raw member data without filtering restricted fields, unauthenticated visitors can retrieve private emails, phone numbers, and other members-only profile attributes. The vulnerability carries a CVSS 3.1 score of 5.3 (Medium) and is classified under CWE-284 and CWE-200, relating to improper access control and information exposure. Minh recommends immediate remediation through REST endpoint permission hardening, field-level privacy checks, and sanitization of JSON API responses.

Mohan Yadav’s ₹10 lakh crore bet on MP

Gold gets its glitter back as prices level out

Model Routing Emerges as Critical Infrastructure Layer in Multi-Model AI Era

As companies scale AI from experiments to production systems, managing multiple specialized models efficiently has become a core engineering challenge. Rather than relying on a single model API, modern AI applications increasingly require a routing layer that directs requests to the most appropriate model based on cost, speed, sensitivity, and complexity. Stripe's acquisition of OpenRouter is seen as a signal that model routing is maturing into a strategic infrastructure category, similar to how cloud computing abstracted physical server management. A routing system acts as an optimization engine, balancing intelligence, latency, compliance, and inference costs across a portfolio of models. Analysts suggest the future AI stack will prioritize task-optimal model selection over simply choosing the cheapest or most powerful option available.

Delhi Police Nabs 72-Year-Old Fugitive in 2008 Extortion Case After 15 Years

Delhi Police has arrested 72-year-old Hari Singh Shekhawat in connection with an extortion case that originated in 2008. Shekhawat had allegedly attempted to extort money from a multinational garment company where he was employed at the time. Authorities pursued the case for approximately 15 years before finally apprehending the accused. The arrest brings a long-pending case to a significant milestone after one and a half decades of investigation.

MCP Standard Lets AI Agents Clip and Edit Video Without Leaving Your Chat Tool

The Model Context Protocol (MCP) has enabled AI agents like Claude and ChatGPT to perform video clipping tasks directly within conversational interfaces, eliminating the need to switch between apps or dashboards. MCP works by giving AI agents a standardised way to call external tools, read available functions, and return structured results — making end-to-end video workflows fully agent-driven. Several video clipping platforms, including OpusClip, Reap, Submagic, and Whipscribe, now expose MCP endpoints, suggesting the feature is becoming a baseline rather than a differentiator among developer-focused tools. Implementations vary significantly in authentication methods, pricing models, and access tiers — some platforms meter agent usage separately, while others bundle it into existing plans. The author, who is building a video tool called Katto with native MCP support, notes that agent access is increasingly being treated as a standard feature rather than a premium add-on.

How Poor AWS Lambda S3 Trigger Design Silently Drops Production Data

A client's data ingestion pipeline appeared stable for eight months until larger batch uploads exposed a critical flaw: roughly 3% of objects vanished without any error logs or alerts. The root cause was an absence of intentional error handling for AWS Lambda S3 triggers, which use an asynchronous, at-least-once delivery model with only two automatic retries by default. Without a Dead Letter Queue configured, any event that exhausts its retries is silently discarded, leaving no trace in CloudWatch. Common mistakes compounding the risk include missing idempotency checks, recursive invocation loops from writing outputs back to the same S3 prefix, and handlers that fail an entire batch when a single record is malformed. Developers are advised to configure DLQs or failure destinations, implement deduplication logic, set reserved concurrency limits, and process each S3 record individually to prevent silent data loss at scale.

Adobe Engineer Beaten to Death by Wife's Relatives in Delhi; Four Detained

A software engineer employed at Adobe was allegedly killed by his wife's relatives following a dispute at his Delhi residence. The victim's family has alleged that the attack was premeditated and a deliberate conspiracy to murder him. The assault reportedly took place after the wife called her family members to the home. Police have detained four suspects in connection with the case. Authorities are investigating under sections pertaining to murder and common intention.

JD Vance endorses Vivek Ramaswamy for Ohio governor ahead of November race

US Vice President JD Vance has publicly endorsed his friend Vivek Ramaswamy for the position of Ohio governor. Ramaswamy, an Indian-American entrepreneur, secured the Republican primary nomination with the support of former President Donald Trump. He is set to face Democratic candidate Amy Acton in the November general election. Vance, who emphasized his Rust Belt background, made the endorsement while campaigning for Republican candidates. The Ohio gubernatorial race is considered a significant contest in the upcoming midterm elections.

Local MCP Server Lets AI Models Review and Comment Directly on Figma Designs

A local Figma MCP Server now allows AI assistants such as Claude, GPT, and Gemini to access and review Figma design files directly within the canvas. The server integrates with MCP-compatible clients including GitHub Copilot in VS Code, Claude Desktop, and Cursor, enabling AI models to analyze layouts, spacing, contrast, and design system consistency. Feedback is pinned as comments to the specific frames being reviewed, and existing comment threads can be read and prioritized by urgency. Because the server runs locally on the user's machine, the Figma personal access token never leaves their control. Setup requires an MCP-compatible AI client, the Bun runtime, and a Figma personal access token with appropriate file and user scopes.

Gujarat liquor tragedy kills 9, leaves 3 critical

New SIM Card Rules on Connection Limits Take Effect from August 24

India's updated SIM card regulations will come into force on August 24, placing stricter controls on the number of mobile connections a subscriber can hold. Telecom operators will be required to identify and notify customers who exceed the permitted limit of connections. Most subscribers across India are allowed up to nine connections, while those in certain regions are capped at six. Initially, verification will be carried out after activation, with a transition to real-time checks planned by November. The new rules are designed to stop individuals from acquiring more mobile connections than the legally prescribed limit.

AiSec Studio Uses Deterministic Pipeline to Stop AI Security Scanners from Hallucinating

Developer AiSec Studio is an offline AI security research platform designed to prevent the hallucination problem common in LLM-based vulnerability scanners. The tool runs every scan through a structured pipeline — parser, rule engine, and knowledge graph — before a local language model ever processes any data. Vulnerability detection is handled entirely by deterministic rules and AST parsers, while the LLM is restricted to explaining and correlating findings already confirmed by the system. All findings must include evidence, a confidence score, a risk level, and a suggested fix, or the system flags them for manual review instead of generating a conclusion. The platform runs fully offline using models like Qwen2.5-Coder or DeepSeek-Coder via Ollama or llama.cpp, with no cloud API involvement at any stage.

Why navigator.sendBeacon beats fetch for analytics on page exit

When a browser tab is closed, fetch requests fired inside a beforeunload handler are often cancelled mid-flight because the browser tears down the page context before the network round-trip completes. navigator.sendBeacon was purpose-built for this scenario, queuing the request and guaranteeing delivery even after the page unloads. Unlike fetch with keepalive, sendBeacon carries no payload-reliability caveats and works consistently across modern browsers. On mobile devices, beforeunload itself is unreliable, as iOS Safari may not fire it when users switch apps or lock their phones. Developers are advised to use the visibilitychange event combined with sendBeacon to reliably flush batched analytics whenever a page enters a hidden state.