What we know
1 model because it is considered too insecure. The company cited security concerns that involve trade-offs similar to those observed in existing public AI models, where improving performance can sometimes introduce vulnerabilities.
Why it matters
This decision highlights the ongoing challenge in AI development of balancing advanced capabilities with strong security protections. Ensuring AI models are safe and secure is critical to preventing misuse and protecting users. 1 underscores the importance of addressing security risks before releasing powerful AI systems.
What is still unknown
1 has not been detailed. It remains unclear what exact trade-offs led to the decision and whether future versions will overcome these issues. No information has been provided about potential timelines for a secure successor or related updates.