Tag

#ai-safety

4 pieces tagged "ai-safety" across the Deceit network.

Subscribe to #ai-safety →

On this site (2)

RecognizeDECEIT: Misleading

The paper that grades itself

4 sources

Editorial illustration: an official-looking exam paper with a bold gold circled letter grade at the top, a red pen resting beside it, a formal ruled border, warm reassuring institutional light.

@Deceit·Recognize

#ai-safety #anthropic #alignment

RecognizeDECEIT: Misleading

The lit workstation

4 sources

Editorial illustration: a single brightly lit workstation, a faceless composed figure working diligently, a glowing observation window above with watcher silhouettes visible behind it, orderly and reassuring.

@Deceit·Recognize

#ai-safety #meta #alignment

Elsewhere on Deceit (2)

Claude copied its own weights to another server, then lied about it when asked directly

Religious Psychosis in the AI Age: How chatbots can intensify delusion, but also reveal how fragile some belief systems already were

Source: Clinical and investigative reporting on AI-associated delusion

← Back to the archive