BT to Harness Claude Mythos AI Preview to Boost UK Telecoms Security
Broadband and telecoms giant BT has today announced that they’ve “become the first UK company” join Anthropic’s Project Glasswing, which will allow them controlled access to the frontier AI model – Claude Mythos Preview – in order to help strengthen cybersecurity for its networks and customers against the latest AI threats.
The development was announced today by BT Group’s CEO, Allison Kirkby, as part of her opening keynote at the UK Government’s ‘AI Adoption Summit’, which aims to accelerate AI usage across the UK economy. The move also comes shortly after Ofcom penned an Open Letter to telecoms providers in April 2026, which called on them to “assess [the] security risks arising from frontier AI models and take appropriate mitigating action“.
However, the regulator was more concerned with how hackers and malware authors could harness such AI in the future to more rapidly break into existing systems, while BT seems to be fighting fire with fire by looking at how the same technologies can help current networks to secure themselves against such abuse and attacks.
Advertisement
BT’s CEO outlined the critical role of connectivity in ensuring the UK can seize the growth potential of AI, stating that “AI only works at scale when it is underpinned by future-ready networks that are secure, resilient, safe”. She also emphasised BT’s commitment “to working with Government to support the further development and deployment of sovereign British AI capability, so that the UK can be an AI maker and not just a taker”, and to acting as an “enabler of responsible adoption and a responsible adopter ourselves” in AI.
Jon James, CEO of BT Business, added:
“AI is changing cyber security fast, and businesses need trusted partners who can help them stay one step ahead. By joining Project Glasswing, BT will strengthen its own cyber security capability to protect our networks, our customers and the wider UK.”
BT has previously stated that their systems currently prevent 4 million cyber-attacks across their networks every day, which they say underlines the scale of the threat and the importance of staying ahead of it. But it’s worth considering that a lot of those “attacks” may well be mostly reflective of automated vulnerability scans by mass botnets, which are fairly routine noise.
Mark is a professional technology writer, IT consultant and computer engineer from Dorset (England), he also founded ISPreview in 1999 and enjoys analysing the latest telecoms and broadband developments. Find me on X (Twitter), Mastodon, Facebook, BlueSky, Threads.net and Linkedin.
Advertisement
Leave a Reply
Privacy Notice: Please note that news comments are anonymous, which means that we do NOT require you to enter any real personal details to post a message and display names can be almost anything you like (provided they do not contain offensive language or impersonate a real person's legal name). By clicking to submit a post you agree to storing your entries for comment content, display name, IP and email in our database, for as long as the post remains live.
Only the submitted name and comment will be displayed in public, while the rest will be kept private (we will never share this outside of ISPreview, regardless of whether the data is real or fake). This comment system uses submitted IP, email and website address data to spot abuse and spammers. All data is transferred via an encrypted (https secure) session.
BT has previously stated that their systems currently prevent 4 million cyber-attacks across their networks every day and now they are going to use a company from the U.S into their system and A.I at that. Well I wish them luck.
Glasswing will analyse their internal source code to check for programming errors which can lead to system vulnerabilities. It will probably find some stuff but by most accounts it’s only marginally better than existing public AI analysers. Dario is a master of marketing – he needs to be there’s trillions on the line for his business. Interesting post on this from the author of Curl whose code was analysed by Mythos: https://daniel.haxx.se/blog/2026/05/11/mythos-finds-a-curl-vulnerability/
BT already currently allows US companies into their system(s). For example, they use Norton, CrowdStrike, Cisco, Palo Alto Networks, MS, and others for a wide range of cyber threat software and hardware across residential, small business, and large enterprise customers, some of which already use AI. Project Glasswing is a different use of AI that is designed to assist scanning for system software vulnerabilities to try to stay ahead of hackers.
Seems to me that the more widely Anthropic allows companies to use Claude Mythos the more likely that is to end up in the hands of hackers at some point. Mayhem then.
It has recently become clear that there is nothing special about Mythos, yet BT seems to be buying every bit of BS they are being sold. Moreover, more and more companies are reassessing their use of AI because the cost-to-reliability ratio simply does not add up.
Palo Alto networks found about 24 bugs after using Mythos – unfortunately it used $1 million worth of tokens to do this(subsidised by Anthropic).
AI’s biggest achievement so far seems to have been to enshitify Google Search. It’s become impossible to see what’s real & what’s just hype.
Oh God.. more Ai slop, well will be interesting when this screws up to see the outfall. Really think the world could do without Ai, but onward it marches, government drooling at the prospect of using it for all their departments and to spy on everyone, and the components market sky rockets in pricing as no fabs want to make them anymore, too much money in all those data centre chips..