09/01/2026
A roundup of what's moving in network security, infrastructure, and the industry. Here's what we're reading this week.
🔹 Data breach exposes up to 14.2 million email logins at six ISPs - BleepingComputer
Japanese operator KDDI disclosed that attackers exploited a vulnerability in third-party software to access an email system shared across five ISPs, potentially exposing the credentials of up to 14.2 million customers. The breach is a good reminder that shared infrastructure multiplies the damage of any single point of failure - and that email credentials, once exposed, tend to end up somewhere unpleasant.
https://www.bleepingcomputer.com/news/security/data-breach-exposes-up-to-142-million-email-logins-at-six-isps
🔹 Modernizing legacy IT with AI without triggering regulatory risk - CIO
This piece's central argument is that AI can accelerate code translation from legacy systems like COBOL, but the real risk lies in undocumented business logic - rules that exist only in the code and the memory of a now-retired analyst, which a model will fill in with whatever seems statistically correct. Worth reading for the practical governance framework, particularly if your organisation is anywhere near a modernisation project.
https://www.cio.com/article/4193445/modernizing-legacy-it-with-ai-without-increasing-regulatory-risk.html
🔹DRC plans $1.5B fiber-optic project along Congo River - Connecting Africa News
The DRC and China's Genew Technologies have signed an MoU to build approximately 2,100km of fibre infrastructure along the Congo River, using the river and its tributaries as deployment corridors to sidestep the country's land-based infrastructure challenges. A practical piece of network geography that illustrates how physical constraints shape connectivity strategy just as much as investment or policy.
https://www.connectingafrica.com/fiber-networking/drc-plans-1-5b-fiber-optic-project-along-congo-river
🔹American 5G is not up to scratch for AI – Ookla - Light Reading
Ookla's research places the US at the bottom of a 22-country table for uplink throughput allocation, and 20th for latency - both below the thresholds needed for emerging AI applications like AR glasses, which are expected to shift network traffic from a 90/10 downlink-uplink split closer to 60/40. The culprit, according to the piece, is heavy reliance on TDD spectrum, which forces uplink and downlink to share the same channel.
https://www.lightreading.com/5g/american-5g-is-not-up-to-scratch-for-ai-ookla
Like these takeaways?
Subscribe to our newsletter for weekly reads, plus tools and tactics to keep your network resilient: https://www.inkbridgenetworks.com/newsletter
