Re: Feedback Request – Embedded Child Protection Solution for Telcos (Academic Research)

[Brian Knight via NANOG <nanog () lists nanog org>]

On 2025-06-24 09:27, evabouchard38--- via NANOG wrote:
> Hi all,
>
> I'm part of a postgraduate team at Dublin City University working with 
> Chirp, a startup developing real-time, embedded child-protection 
> software for telecom operators. The solution analyzes data traffic on 
> children’s devices to block harmful content and alert parents to risks 
> such as grooming, cyberbullying, or self-harm — all while respecting 
> privacy and working natively within telco infrastructure.
>
> As part of our MSc practicum, we’re seeking feedback from telecom and 
> network professionals on the commercial, technical, and regulatory 
> feasibility of such an approach.
>
> Would you be open to completing a short, 10-minute questionnaire?
>
> 🔗 https://dcusurveys.qualtrics.com/jfe/form/SV_8oBhWiZMRrUh1zM

From the content of the survey and your website, your team seems well 
aware of where your app sits in the network stack: on the user's device, 
not in telco space.

Honestly, the survey reads a bit like a sales pitch to residential ISP 
product managers. I'm mildly curious what they would think of offering 
this app.

Does your solution require equipment deployed on the ISP network?

> We’d be very grateful for your insights. Happy to follow up with more 
> technical or contextual details if helpful.

Personally, I haven't found any value in my home ISP's value-added 
services (like Norton Anti-Virus). But that would certainly vary by 
locale.

Also, historically, I know pattern matching software works well when the 
content to be filtered is relatively static, like malware. Pattern 
matching works less well with images, audio, and human-generated text.

As an engineer, I would want a more thorough description of your 
filtering strategies to understand why your app might work where others 
have failed. Real-world false-positive and false-negative rates would be 
key if I were a parent. Any LLM integration would also need to be 
detailed, to understand where a child's data might be shipped for 
analysis.

> Thanks in advance for your time!
>
> Best regards,
> Eva Bouchard

Thanks,

-Brian
_______________________________________________
NANOG mailing list 
https://lists.nanog.org/archives/list/nanog () lists nanog org/message/JUVZ7LIPREOMYS6BRZ5W335B53BTQKDB/