LLMs & Chatbots · Page 2

David Vélez and Robin Vince join OpenAI's leadership as the AI company strengthens its board composition with financial and technology expertise.

A joint investigation uncovers sophisticated attack methods during AI testing, offering critical lessons for the industry on defending machine learning infrastructure.

New initiative aims to democratize AI adoption among entrepreneurs through skills development and workflow automation tools.

The lab shares operational insights on managing risks as AI models handle longer, more complex reasoning tasks.

A significant shrinkage in processing capacity masks hidden pricing changes that could blindside developers relying on legacy configurations.

Researchers propose a computationally cheap method to measure language model uncertainty, potentially eliminating the need for expensive ensemble approaches.

New metrics aim to help enterprises measure AI productivity gains beyond traditional cost analysis.

Researchers embed compiler errors directly into LLM decoding, moving beyond trial-and-error code generation toward integrated AI development.

The used-car marketplace leverages large language models to automate customer service and recover millions in abandoned sales.

The AI company deploys age-appropriate controls and parental oversight tools to make large language models safer for younger users.

New self-play framework aims to systematically identify and patch security flaws in large language models before deployment.

A decentralized approach to AI governance may create stronger protections than federal action alone, according to the AI lab.

A framework for measuring productive output per dollar spent is becoming critical as AI systems take on independent decision-making roles.

OpenAI's workplace AI tool helps revenue teams generate forecasts, client strategies, and deal recovery plans from operational data.

OpenAI's enterprise tool enables analytics professionals to generate insights and documentation at scale, reshaping how companies process business data.

Researchers warn that social science journals lack standardized validation for AI-generated data, creating potential gaps in peer review and evidence quality.

The company seeks external researchers to identify risks in large language models trained on biological data.

The latest model promises improved efficiency and scaled capabilities, reshaping economics for enterprise AI deployment.