Recent high profile 'attacks' by internal OpenAI agents have the industry and media in a panic. The complex series of problems that make up this wider issue need to be evaluated and solved, with clear heads, and from a number of different angles. This article suggests an approach for solving the bulk of the issues the industry has seen so far.Read article3 min read AI Deepfakes Security Hacking
With the sporadic, yet increasing sound of AI agent foot-guns going off in the distance, I'm left wondering whether we've forgotten the lessons of the past. Or perhaps in the rush to LLMify everything we're forgetting to put up common guard rails. What am I misssing?Read article