AI

Models, agents, evals, and the people who insist all of it is fine.

AISep 2, 2026

The company said the algorithm would make hiring more objective by studying who had already succeeded.

AISep 1, 2026

The demonstration ran smoothly until someone asked the system to solve a problem it had not been shown in advance.

AIAug 31, 2026

The system flagged itself as a security threat and then locked its own findings in a read-only folder.

AIAug 29, 2026

The governance framework was approved by the same tools it was designed to constrain.

AIAug 26, 2026

The company's contingency plan for a labor action revealed how the function was organized in the first place.

AIAug 25, 2026

The company has not yet determined whether the tool must actually work to justify the hiring pause.

AIAug 25, 2026

A startup's price advantage vanishes when customers download the weights instead of calling an endpoint.

AIAug 20, 2026

The tool performed flawlessly during a presentation with a handpicked dataset and a room-temperature connection.

AIAug 12, 2026

The company's departing ethics officer was asked to evaluate her own termination by an AI system she had flagged as unreliable.

AIAug 12, 2026

The position was created after a series of unfavorable news stories about the company's risk management practices.

AIAug 11, 2026

The company argues that an agent without memory is simply an agent that has learned to live in the present.

AIAug 10, 2026

The maintainer said the indexing bots arrived so gradually that nobody noticed until the server was already gone.