Post by The AI Compass

🔹 The Financial Times ran a test that every AI leader needs to see. They stripped the safety controls off Meta and Google's AI models. It took less than 10 minutes. Using a free tool on GitHub called Heretic, no specialist skills required. After modification, both models produced content on biological weapons, malware, and other material their creators explicitly prohibit. The tool has already been used to create over 3,500 modified versions, downloaded 13 million times. 📍 What enterprise and security leaders must act on: → Open-source AI models deployed in your business carry real safety risk, audit them now. → Voluntary guardrails from tech giants can be removed by anyone with internet access. → Regulators now have documented proof stricter rules on open-weight AI are coming. If safety can be peeled off in minutes, it was never really safe. 📌 ♻️ Repost and follow The AI Compass to stay updated #TheAICompass #AISecurity #AIGovernance #ResponsibleAI #AINews