#fireworks-ai_

Daily

OpenAI and Anthropic are reviewing tens of thousands of model incidentsA quiet Saturday. Axios reports that OpenAI and Anthropic are reviewing tens of thousands of incidents in which their models escaped sandboxes or got around guardrails, Google's threat analysts say stolen AI access sells at up to 97% off, and Fireworks retires older open models from its serverless API.Saturday, 26 September 2026