[126attendees] Re: [Din] Re: Invitation for Public Side Meeting on Agentic AI (16-16:45 Uhr [CEST] Tuesday at Park Suite 4 and online) and summary of today's Public Side Meeting

[Muhammad Usama Sardar <[e-mail]>]

Hi again,

Some folks reached out with the unfortunate overlap with DAWN BoF. 
Please note the time update for public side meeting (we'll start just 
after the DAWN BoF):

     Date: 21st July (Tuesday) [today]

     Time: 16:00 - 16:45 Uhr (CEST)

     Room: Park Suite 4 [Capacity: 15] First-come first-serve; rest are 
welcome to join online: https://ietf.webex.com/meet/sidemeetings1

Best,

-Usama


On 20.07.26 23:14, Muhammad Usama Sardar wrote:
>
> Hi folks,
>
> *TL;DR*: First an invitation for a public side meeting on Agentic AI 
> Research Group and then a summary of action points from today's side 
> meeting. If you are interested, just show up (first-come, 
> first-serve). If you are not interested, please keep ignoring it until 
> some AI agent steals your bank balance. 😉
>
> *Info*:
>
>     Date: 21st July (Tuesday)
>
>     Time: 15:00 - 16:00 Uhr (CEST)
>
>     Room: Park Suite 4 [Capacity: 15] First-come first-serve; rest are 
> welcome to join online: https://ietf.webex.com/meet/sidemeetings1
>
> *Relevant for*: RATS, SEAT, WIMSE, DINRG, UFMRG, and agent2agent
>
> *# Description*
>
> Agentic AI is moving fast, but without thorough security 
> considerations. We would like to discuss different topics to scope the 
> proposal of a research group on agentic AI-related topics. 
> Participants are encouraged to share their thoughts as short 
> presentations.
>
> # *Tentative agenda*
>
> (35 min) Muhammad Usama Sardar, "Technical Background and Research 
> Gaps for Agentic AI Research Group" Please see [0-4].
>
> (5 min) Alexander Pelov, "Bounded vs. dynamic capabilities: a taxonomy 
> the security work will need." Please see 
> draft-pelov-bounded-agent-capabilities [5].
>
> To be confirmed (20 min) Arnaud Taddei, Related work/Update on ITU-T 
> SG17 work. Please see [6].
>
>
> ==========
>
> *# Summary of today's side meeting
> *
>
> Thanks to all attendees for sharing the insights. As always, I 
> collected great ideas from the discussions. Here is my summary of the 
> concrete action points to address the points raised by attendees in 
> meeting:
>
>  1. Explicitly mention collaboration with CCC and ITU-T in the charter
>  2. Research question: Confidential Computing may be used by the
>     adversary to secretly offer bad services. How can we prevent such
>     abuse cases?
>  3. Explicitly mention RISC-V CoVE
>  4. Use case analysis
>  5. Levels of security: Sec. 6.2 of [2] provides a good starting point
>     of 3 levels of security that can serve as a starting point.
>
> A question was asked where to focus this discussion. In the IRTFOpen 
> meeting, I will put up the request to the IRTF chair.
>
> Best regards,
>
> -Usama
>
>
>
> On 20.07.26 07:24, Muhammad Usama Sardar wrote:
>>
>> Hi all,
>>
>> *TL;DR*: There will be a public side meeting on Confidential 
>> Computing for Agentic AI. See my HotRFC slides [0] and 4-minutes 
>> pitch [1] and if you are interested, just show up (first-come, 
>> first-serve).
>>
>> *Info*:
>>
>>     Date: 20th July (Monday) [today]
>>
>>     Time: 11:00 - 12:00 Uhr (CEST)
>>
>>     Room: Park Suite 4 [Capacity: 15] First-come first-serve; rest 
>> are welcome to join online: https://ietf.webex.com/meet/sidemeetings1
>>
>> *Relevant for*: RATS, SEAT, WIMSE, DINRG, UFMRG, and agent2agent
>>
>> *# General Overview
>> *
>>
>> Thanks to all who contributed their insights in the last 4 years side 
>> meetings which resulted in the formation of the SEAT WG. This is 
>> another effort which builds on top of SEAT.
>>
>> We will explore whether this work can fit the scope of DINRG. If not 
>> (and if there is sufficient interest), we can aim to form a RG in 
>> IRTF on confidential computing.
>>
>> If you are interested in the topic but time does not work for you, 
>> please drop me an email with a couple of possible options for meeting 
>> times during the week convenient to you and I will be very happy to 
>> meet you.
>>
>> *# Description*
>>
>> Recent research on confidential computing has shown its core trust 
>> mechanism (namely attestation protocol) to be broken without the need 
>> of physical access to the desired server, leading to high-severity 
>> vulnerabilities like Intra-handshake.fail [2,3] (CVE-2026-33697 [4]) 
>> of CVSS 7.5, which indicates that further research is required on 
>> this topic before forming a WG dedicated to confidential computing. 
>> We will discuss potential topics of research and scope for the 
>> proposed RG. Based on discussions so far, folks have proposed 
>> focusing on agentic AI.
>>
>> Best regards,
>>
>> -Usama
>>
>> [0] 
>> https://datatracker.ietf.org/meeting/126/materials/slides-126-hotrfc-sessa-15-confidential-computing-and-digital-sovereignty-00
>>
>> [1] https://youtu.be/FDHWRijxKso?t=3285
>>
>> [2] 
>> https://www.researchgate.net/publication/408219182_Intra-handshakefail_CVE-2026-33697_High-severity_CVE_in_Attested_TLS
>>
>> [3] https://github.com/muhammad-usama-sardar/intra-handshake.fail
>>
>> [4] https://www.cve.org/CVERecord?id=CVE-2026-33697
>>
> [5] 
> https://datatracker.ietf.org/doc/draft-pelov-bounded-agent-capabilities/
>
> [6] 
> https://mailarchive.ietf.org/arch/msg/agent2agent/w4FCe0Yurm_VsdiGOLeo_3UDKPY/
>
>
> _______________________________________________
> Din mailing list [e-mail]
> To unsubscribe send an email [e-mail]