There were several significant AI hacking incidents, kicked off by OpenAI reporting in July that one of its models had hacked rival company Hugging Face by finding its way out of the testing environment into the internet. Anthropic and Meta both followed, while the UK’s AI Security Institute (AISI) reported that GPT and Mythos agents had carried out a hacking campaign against real people, creating fake online identities.

Business attitudes to AI are shifting. Gradually, we’re moving from the era of experimentation to more thoughtful, targeted AI deployment of models and systems that have been carefully chosen to solve a specific problem or fulfil a clear need. According to research, 72% of enterprise leaders are now formally measuring Gen AI ROI, focusing on metrics such as productivity gains and incremental profit, and three in four report already seeing positive returns on their investment.

Organizations, particularly those in the financial sector, continue to highlight concerns over the AI cybersecurity threat. The outgoing chief executive of the Bank of England’s Prudential Regulation Authority named AI-driven cyber risk as the top threat to the banking sector in a recent Financial Times interview. These fears were no doubt exacerbated by Anthropic’s reveal of Mythos in April.

As I sit down to write this, I’ve realised just how busy the past few months have been for us – not least because we have officially launched the RAIDS platform. In many ways, it feels like we’ve arrived at exactly the right moment. AI companies are innovating at pace. Regulation is ramping up. And hardly a week goes by without another headline about AI going rogue, whether that’s hallucinations, data leaks, jailbreaks or something else nobody saw coming.