Cyber security experts on both sides of the Atlantic are raising red flags over the growing AI attack threat. According to the latest State of Information Security Report from IO, the biggest worries right now are shadow AI (unsanctioned use of AI tools) and data poisoning attacks targeting LLM models.
The report polled 3,000 IT security leaders and offers a big-picture snapshot of how AI threats are evolving in real time.
One of the standout findings is that 26% of organisations have already faced a data poisoning attack. These occur when attackers tamper with training data to deliberately change how an AI model behaves.
The danger here is twofold: it can either sabotage businesses that rely heavily on Artificial Intelligence, or it can help attackers bypass defences by making detection systems less effective. What was once seen as a “theoretical” threat is now happening in the wild.
The report also found that 37% of enterprises have employees using generative AI tools without permission. This shadow AI trend raises big issues:
Earlier this year, DeepSeek’s flagship LLM R1 was shown to have multiple vulnerabilities, and the company even exposed a database of sensitive chat histories. It’s a reminder that unapproved tools can create more problems than they solve.
When asked about what they see coming, IT leaders flagged some key concerns:
Interestingly, while deepfake-related incidents fell from 33% last year to 20% this year, it’s still considered a major risk.
Despite all these challenges, most respondents believe they’re prepared. High numbers reported confidence in defending against:
On top of that, 75% of businesses are rolling out acceptable use policies for AI, which should help curb unsanctioned usage.
While it’s great to see that UK businesses feel aware of the threat of a cyber attack, it doesn’t mean we’re as protected against it as we can be. Employee Cyber Security training remains paramount- after all, human error is still the biggest cause of data breaches worldwide.
We hope you’ve liked this blog. Stay tuned for more blogs like this. Stay safe!
