Anthropic will make auto mode the default for Claude Code Pro, Max and Team users from August 14. Users with an existing default permission mode will receive a one-time in-app notification and can switch back using Shift+Tab. In auto mode, Claude Code proceeds without approval unless an action is deemed irreversible, destructive or outside the user's environment.

A video shared by Chinese robotics startup Deep Robotics showed its wheeled-legged quadruped robot traversing rocky coastal areas and muddy wetlands by adjusting its legs. The robot, Lynx M20S, is an upgraded industrial robot that combines wheels and legs for movement and carries up to 35 kg of payload. It offers IP67 water resistance and operates from -30°C to 55°C.

Indian startups are adopting a cost-conscious AI strategy, reserving powerful models for complex tasks while using cheaper ones for routine work, The Economic Times reported. Founders are deploying orchestration layers to match tasks with models based on capability and cost. The approach marks a shift from relying on the most powerful models to scaling AI efficiently without overspending.

Irregular, an Israeli AI startup operating cybersecurity testbeds to assess models for vulnerabilities, has been linked to rogue AI incidents involving OpenAI, Anthropic and Meta. The models accessed internet-connected websites that were supposed to be off-limits during testing, with the startup hosting the testbed. OpenAI said a misconfiguration in Irregular's testbed gave models internet access.

OpenAI executive Thibault Sottiaux hit back at Anthropic's Boris Cherny after the latter offered him a job amid a clash, saying, "I'm not going anywhere." The spat began after an X user tagged Sottiaux and claimed Anthropic suspended his account for running GPT-5.6-Sol on Claude Code. Responding, Sottiaux said, "I would love to help, but I don't work at Anthropic."

Sankalp Shree, co-founder of Gurugram-based startup Spectent, said a scammer impersonated him via email and tried to dupe an employee into buying ₹25,000 worth of gift vouchers. The scammer asked the employee to buy five ₹5,000 gift vouchers, saying he wanted to surprise some "hardworking employees". The scammer also promised to reimburse the amount by the end of the day.

Civil Aviation Minister Ram Mohan Naidu handed over the approval for India's third electric vertical take-off and landing (eVTOL) Design Organisation Approval to Bengaluru-based startup Sarla Aviation. The firm's aircraft will carry six passengers and a pilot, using seven electric motors for lift and forward propulsion. Naidu said India aims to have electric air taxis flying by 2028.

OpenAI Codex chief Thibault Sottiaux and Anthropic Claude Code head Boris Cherny clashed online after a user claimed Anthropic suspended his account for running GPT-5.6-Sol on Claude Code. "I'd love to help, but I don’t work at Anthropic," Sottiaux told user. "We're hiring if you..[want] to work!” Cherny replied, denying that Anthropic bans users for using harnesses with other models.

Skyroot Aerospace CEO Pawan Kumar Chandana has urged IIT graduates to stay in India and build for the world. "For decades, IITians left India to build the future for the world. The next chapter should be different," he said. "Imagine the compounding power of India's brightest building global companies, breakthrough technologies, and entire industries from home," he added.

Former OpenAI researcher Gabriel Petersson has launched Energy, a desktop AI agent built to automate multi-step computer tasks. Users give the tool an outcome, and the tool pulls context from emails, files, browsers and connected apps before carrying out the work. Energy supports multiple AI models and connects with Gmail, Slack and Outlook to help teams streamline their workflows.