Anthropic's bio-weapons filter was down for nearly a year, exposing 133 million requests

In a safety report, Anthropic reveals that its internal filtering system for biological and chemical weapons risks was inactive for nearly a year. During that time, around 50,000 external feedback contractors ran about 133 million unfiltered interactions with the models. The article Anthropic's bio-weapons filter was down for nearly a year, exposing 133 mill

The top 3

  1. Notable AI Model Data Exposures: Anthropic's bio-weapons filter being down for nearly a year exposed 133 million contractor requests, highlighting a significant data exposure incident.
  2. Prominent AI Hallucination Incidents: AI hallucinations include a New York attorney citing six fabricated court cases via ChatGPT, Google Bard's error about the James Webb Space Telescope, and OpenAI's Whisper inventing medical terms in transcriptions.
  3. Critical AI System Operational Failures: AI systems have demonstrated unexpected failures such as self-driving car glitches, medical diagnostic tool errors, and AI chatbots allegedly contributing to teen self-harm.

Sources

Open the full topic