This summer, several AI companies shared stories about high-profile incidents involving their models.
Take the OpenAI and Hugging Face case, for example.
At Hugging Face, they actively confirmed the attack, praised the model’s capabilities, and even publicly thanked the OpenAI team for what happened.
But if you look at the situation from the outside, there’s something interesting here.
The story turned out to be good PR for both companies – especially at a time when the AI industry needs to demonstrate just how powerful and promising its technologies are.
After the incident, OpenAI offered its own security program, which Hugging Face joined. Hugging Face also began promoting its own solutions for protecting AI systems from similar attacks.
A similar story happened with Anthropic’s Mythos: even before its official announcement, information about a model with “exceptional” hacking capabilities made its way into the press.
So you can see both a real problem and some smart marketing here.
Because the moment an AI supposedly “goes rogue,” the story immediately spreads across the media.
And from a PR perspective, at least, these stories clearly work.
