Ofcom has urged operators including BT, Vodafone and Virgin Media O2 to be cautious about how they deploy the technology amid concerns it could trigger outages if it goes wrong.

Telecoms companies are increasingly using AI to monitor and manage their networks. This can allow them to detect and repair issues more quickly, improving the reliability of connections. But the technology carries the risk of serious outages if it fails.

Ofcom said: “Network automation carries risk, potentially of catastrophic network failure, so it is crucial that network automation is very carefully considered in every aspect.”

The warning comes amid growing concerns about the security risks linked to AI, in particular the emergence of so-called “frontier” models that make it easier for hackers to identify and exploit vulnerabilities.

Earlier this month BT became the first British company to confirm its participation in a trial scheme of Anthropic’s Claude Mythos Preview – an AI model deemed too dangerous to be released publicly.

But the move proved short-lived, as Donald Trump quickly intervened to block foreign nationals from using the technology, forcing Anthropic to suspend access for all users.

Telecoms providers are considered particularly exposed to attacks by hostile actors given their role in providing critical national infrastructure.

Natalie Black, Ofcom’s head of infrastructure and connectivity, wrote to operators in April urging them to assess the security risks from frontier AI models and take appropriate action.

She said: “Frontier models will continue to advance quickly, increasing both the scale and speed of cyber threats. Continued and proactive investment in cyber defence is critical.”

But the regulator’s latest guidance underscores the risks related to AI even without interference from hostile actors.

It also raises the prospect of fines if telecoms companies fail to manage the risks associated with the new technology.

Under the Telecoms Security Act, operators are obliged to take “appropriate and proportionate” measures on security, such as ensuring their networks are designed to reduce the risk of cyber attacks and to limit the impact of a breach if it does occur.

Ofcom fined BT £17.5m for a “catastrophic” 10.5-hour blackout of the 999 emergency network in 2023 that affected nearly 14,000 calls to police, fire, ambulance and coastguard services.

Telecoms bosses argue that the industry already faces stringent rules around resilience while also being asked to lower prices and improve coverage, adding that tech companies do not face the same regulatory burden.

Mobile UK, an industry group, has urged the Government to provide public funding for these improved resilience and security requirements, arguing that it should no longer be primarily the responsibility of the private sector.

An Ofcom spokesman said: “Telecoms networks have long managed risks stemming from AI. However, advances in technology – especially in frontier AI – are increasing these risks.

“Companies must take appropriate steps to reduce the risks of security compromises occurring. We wrote to firms earlier this year about their responsibilities, and we’re keeping a close eye on them to ensure compliance.”

A BT spokesman said: “We have been using automation in the operation and support of our networks for many years to deliver products and services at scale and continue to take a considered approach to where automation is applied.

“BT has comprehensive controls and governance in place to manage how automation is deployed across our networks.

“Alongside this, we’re continuing to explore emerging technologies that can further strengthen network performance in critical failure scenarios, while also enhancing our cybersecurity capabilities.”

Source link