Four in five (81%) UK CISOs believe DeepSeek — a China-based AI chatbot raising global security concerns due to its data handling practices and vulnerability to misuse — must be urgently regulated by the UK Government before it sparks a full-scale national cyber crisis, according to Absolute Security’s UK Resilience Risk Index Report.

In response to these growing risks, over a third (34 per cent) have implemented full bans on AI due to cyber security concerns, while 30 per cent of CISOs say they’ve already pulled the plug on AI tools within their organisation.

The findings come from a recent survey commissioned by Absolute Security, which polled 250 UK CISOs at enterprise organisations via independent research agency Censuswide, to assess how businesses are coping with accelerating cyber challenges in an increasingly AI-powered world.

DeepSeek, the rapidly rising new cutting-edge AI platform, has raised significant cybersecurity concerns due to its potential to expose sensitive data and be misused by cybercriminals. These issues have led organisations and governments to reconsider their cybersecurity strategies.

Businesses are already struggling to cope with the increasing complexity of cyber threats, as shown by the recent Harrods breach. As CISOs grapple with this evolving landscape, the added layer of AI-powered threats is prompting a re-evaluation of cyber defences.

Three out of five (60 per cent) UK CISOs now predict a rise in cyber-attacks as a direct result of DeepSeek, with another 60% saying this AI technology is already complicating privacy and governance frameworks, making their job more difficult.

These concerns reflect a clear shift in mindset, with 42% of CISOs now seeing AI as a bigger threat than a help to cybersecurity.

The readiness gap is just as concerning, with nearly half (46%) of security leaders admitting their teams are not prepared to handle AI-driven threats, such as those posed by tools like DeepSeek. The rapid development of DeepSeek is outpacing their defences, according to survey findings, creating a growing risk that many believe can only be managed through government regulation.

“Our research highlights the significant risks posed by emerging AI tools like DeepSeek, which are rapidly reshaping the cyber threat landscape,” Andy Ward, SVP International of Absolute Security commented.

Recommended reading

• US CISOs Back AI Deregulation, Except for DeepSeek
• Beyond DeepSeek: 3 Critical Questions For The Future of AI
• South Korea Suspends DeepSeek Over Privacy Concerns

“As concerns grow over their potential to accelerate attacks and compromise sensitive data, organisations must act now to strengthen their cyber resilience and adapt security frameworks to keep pace with these AI-driven threats. That’s why four in five UK CISOs are urgently calling for government regulation. They’ve witnessed how quickly this technology is advancing and how easily it can outpace existing cybersecurity defences.

“These are not hypothetical risks. The fact that organisations are already banning AI tools outright and rethinking their security strategies in response to the risks posed by LLMs like DeepSeek demonstrates the urgency of the situation.

“Without a national regulatory framework—one that sets clear guidelines for how these tools are deployed, governed, and monitored—we risk widespread disruption across every sector of the UK economy. The time for debate is over. We need immediate action, policy, and oversight to ensure AI remains a force for progress, not a catalyst for crisis.”

Despite the risks, investment in AI talent is accelerating. Eighty-four per cent of organisations are prioritising the hiring of AI specialists in 2025, and 80 per cent have committed to AI training at the C-suite level, hoping that upskilling AI talent can outweigh any increasing threats.

Most companies don’t intend to retreat from AI—they want to face it head-on. To use AI safely, CISOs say they need clear rules, stronger government oversight, a skilled AI workforce, and a national plan to deal with the specific risks of DeepSeek and similar tools.