GPT-5.4 is also more reliable, producing 18% fewer errors and 33% fewer false claims than GPT-5.2, according to OpenAI.

Vibe coding isn’t just prompting. Learn how to manage context windows, troubleshoot smarter, and build an AI Overview extractor step by step.

The company cited attorney-client privilege in rejecting a bid for information related to its deletion of two datasets used to train old versions of its AI system. By Winston Cho It was a highly ...

Pillar Security discovers two critical vulnerabilities (CVSS 10.0) in popular workflow automation platform affecting hundreds of thousands of deployments – enabling attackers to decrypt stored ...

A Chrome extension posing as an AI assistant exposed more than 10,000 users, secretly harvesting OpenAI API keys and sending data to attacker-controlled servers. Researchers say at least 459 API keys ...

OpenAI has reorganized some of its leadership and picked a familiar face to lead its push into selling AI to business customers as the company looks to catch up to its rivals in 2026. The company ...

Michelle is Lifehacker's Associate Tech Editor, and has been reviewing games, laptops, phones, and more for over 10 years. She is based in New York City and holds a master's degree from NYU. January ...

Lawyers representing the Daily News and an array of news organizations suing OpenAI for allegedly stealing and distorting their reporters’ work have asked a Manhattan judge to sanction ChatGPT’s ...

Jerry Tworek, a vice president of research at OpenAI and a key figure in the company's work on AI reasoning models, is leaving the company after nearly seven years. Tworek announced his departure in a ...