OpenAI secretly funded and had access to a benchmarking dataset, raising questions about high scores achieved by its new o3 ...

AI might excel at certain tasks like coding or generating a podcast. But it struggles to pass a high-level history exam, a ...

OpenAI just pulled a Theranos with o3 by claiming record-breaking performance on the FrontierMath benchmark while having ...

The company developed DeepSeek-R1 by using pure reinforcement learning on top of DeepSeek-V3-Base, and matched or beat o1 on some benchmarks.

A coalition of news organizations led by The New York Times claim the exploitation of their online news stories to train ...

OpenAI has quietly removed language endorsing 'politically unbiased' AI from one of its recently published policy documents.

Red teaming has become the go-to technique for iteratively testing AI models to simulate diverse, lethal, unpredictable attacks.

OpenAI has cut off a developer who built a device that could respond to ChatGPT queries to aim and fire an automated rifle.

Jack Clark was a tech reporter at Bloomberg and I was his editor. I didn't want him to leave. He left anyway.

In May, OpenAI’s superintelligence safety team was disbanded and several senior personnel left due to the concern that ...

A trove of newly released documents reveals Meta’s plans to use book piracy site LibGen to train its AI models.

OpenAI has released an "economic blueprint" for AI, detailing policies it thinks can be implemented with the US government ...