MWC 2026: Deutsche Telekom leads telco peers in showcasing AI security
Report discovered via Awesome Annual Security Reports.
Summary
A HardenStance briefing finds Deutsche Telekom presented the clearest AI-security vision at MWC 2026, including controls for AI agents. Telefónica also emphasized security, while Orange showed little evidence of AI-risk mitigation.
Key points
- Deutsche Telekom presented a five-stage AI-security roadmap covering governance, workflow enforcement, network defense, and AI identity.
- The briefing says AI agents should be secured like other software: MCP and A2A lack built-in security, and agents need identities and least-privilege permissions.
- Deutsche Telekom described using AI agents for chatbot red teaming and said it has deployed 6,000 honeypots to gather data on attacks targeting LLMs.
- Deutsche Telekom and Google Cloud demonstrated MINDR, a multi-agent system for autonomous diagnostics and operations across telecom networks.
- Deutsche Telekom plans to test its Mobile.ID secure chip with employees in spring 2026, with a commercial launch targeted for summer.
- The briefing says Telefónica presented AI-security themes but Orange offered little evidence of AI-risk mitigation at its MWC booth.
- The report notes South Korean breach impacts, U.S. telcos’ accountability over Salt Typhoon breaches, and Telia Norway’s disclosure of another sizable breach during MWC.
Article Details
- Event Type
- MWC26 telecommunications cybersecurity assessment, including discussion of recent breach disclosures and telcos’ AI security strategies.
- Impact
- The briefing reports that Telia Norway disclosed a sizable data breach during MWC and refers to significant breach impacts in South Korea and Salt Typhoon breaches in the United States; it gives no further details about the affected data or operational consequences. The author assesses Deutsche Telekom’s AI security vision favorably, describes Telefonica’s security focus as less prominent, and says Orange showed little evidence of AI risk mitigation at MWC.
People
Christel HeydemannOrange CEO; discussed the opportunities and risks of agentic AI in her MWC keynote.Ollie WhitehouseIdentified as CTO of the UK’s National Cyber Security Centre; awarded the telecom sector a C+ cybersecurity grade at MWC 2025.Patrick DoneganHardenStance founder and principal analyst; author of the briefing.Stefan SchneiderIdentified as Deutsche Telekom’s Head of AI Security.Summer YueThe briefing reports that her OpenClaw AI agent began deleting her email inbox despite her instructions.Thomas TschersichIdentified as Deutsche Telekom’s Chief Security Officer and CEO of Telekom Security; discussed verification and trust in the AI era.
Threat Actors
Vendors
AnthropicAnthropic resisted U.S government demands to shelve its ethics – so OpenDeutsche TelekomCongress saw Deutsche Telekom excel among its European telco peers with a visionGoogle Cloudnetwork domain at MWC. In partnership with Google Cloud, DT demonstrated Multi-OpenAIcompetitors, OpenAI and xAI, exploited the fallout by hurriedly signing new dealsOrangebrand new “Trust in the Future” branding, Orange had very little to say orPalo Alto Networksincluding Palo Alto Networks, in evolving its platform roadmap.TelefónicaTelefonica also focused more than usual on trust and security this year. Despite itsxAIAI and XAI stepped in. Two days before MWC opened its doors, the U.S and Israel
Products
Claude AIthe owner of Claude AI and a key supplier to the military. Around the time of thoseMagenta Security Mobile.IDlaunch within the next six months or so. Magenta Security Mobile.ID is a secure chipMINDRAgentic Intelligent Network Diagnostics & Remediation (MINDR), a multi-agentic AIOpenClaw▪ OpenClaw ran amok the week before MWC: Days before MWC, Summer Yue,RentAHumanSecure on the last day of MWC, RentAHuman is a new platform that “lets AI agents