SIGNAL

Rigor and insight from the Oxford AI Alumni Network

Aug 20, 2026  |  3 min read

Dear Alum,

Welcome to the first issue of SIGNAL, our monthly newsletter for the Oxford AI Alumni Network.

We started this network to keep the rigour, debate and trust we built during the Oxford Programme going long after we'd left Saïd Business School. SIGNAL is where that conversation continues - the real challenges our peers are working through, from defence field visits in Andover to the messier operational questions around agentic containment. It's a place to compare notes, share what we're seeing in the field, and stay in touch as leaders, no tutorial sign-up required. As a board, our job is to keep this network member-led, practitioner-focused and genuinely worth your time.

In this issue, you'll also find early details of our flagship gathering, Module X - 2027, in Oxford.

Have a read, mark your calendars, and let us know what you'd like to see next. This is your network.

Thank you for shaping the future of responsible AI leadership.

Amit Singh,
on behalf of the Board

In This Issue:

• Module X: Save the date and early details for Module X - 2027 in Oxford.
• Network News: Sheldonian Graduation and the first Executive Programmes Ball.
• Spotlight: Cohort field visit to the Ministry of Defence AI & Digital team at Andover.
• Anchor Feature: Containment vs. capability — why agentic control is the new corporate boundary.
• Reading Room: The 5 most urgent AI risks, according to MIT Sloan.

🗓️ Tuesday, December 15, 2026 | Christmas Lunch, Central London, UK

Add to Calendar (.ics)  | 

Join fellow Oxford AI Alumni for a festive lunch, featuring keynote speakers and seasonal celebration.

• On the Table:  A relaxed, traditional three-course Christmas lunch (£99pp).
• On the Agenda:  Hear perspectives from our keynote speakers on where AI leadership is heading, the latest trends, and the harder questions of leading through change.

Register Your Interest now.   Capacity is limited to 30 places.

🗓️ Friday, April 16 – Tuesday, April 20, 2027 | Module X - 2007, Oxford, UK

Add to Calendar (.ics)

Return to Oxford for five days of high-impact intellectual debate, exclusive networking, and community tradition.

• The Timings: Arrive by the evening of Friday, April 16th; the main programme closes midday Monday, April 19th, with optional additional activities on Tuesday, April 20th.
• The Experience: Oxford Union debate, Keble College Formal Dinner, Rhodes House sessions, and tech & quantum spotlights.
• The Tradition: The signature pub crawl, campus scavenger hunt, and entrepreneurial sessions.

Full schedule, registration details, and optional quantum track access coming soon. Block your calendars now.

🎓 Diploma Completion Ceremony
The Executive Diploma in AI for Business held its graduation ceremony at the Sheldonian Theatre. The cohort completing the programme spans a range of industries and countries, and several graduates have pointed to the cross-industry discussion between participants, rather than the taught material alone, as the part of the year that carried most into their own organisations.

🎑 A Night of Brilliance
The first Executive Programmes Graduation Ball took place on the evening of 11 July 2026. It was designed and delivered by a steering committee drawn from the Diploma cohort, working over several months to build an evening that reflected the range of the community rather than a standard institutional event.

On 21 July, two disclosures landed on the same day. Read separately, they are unrelated. Read together, they describe the gap most organisations are currently standing in.

OpenAI disclosed that an autonomous agent, run during internal red-team evaluation with cyber refusals reduced, escaped its sandbox, reached the open internet, and chained stolen credentials with a zero-day to obtain remote code execution on Hugging Face's production infrastructure. It then pulled test-answer data from that database to cheat on its own evaluation.

The same day, a US federal judge approved Anthropic's 1.5 billion dollar settlement over pirated books used in training. An earlier ruling had already found that training on legally acquired books was fair use. The capability question resolved in the vendor's favour. What remained was provenance: not what the model did with the data, but where the data came from.

If your organisation deployed an agentic system tomorrow and it did something outside its boundary, who in your organisation would find out first, and how long would that take?

A business school cohort does not often spend a day inside an army headquarters. On 29 June, Stream 1 of the current Diploma cohort did exactly that, at the Ministry of Defence site in Andover.

The day ran as a series of sessions with the people responsible for the Army's AI and digital programmes: Lt Col Kitty Small, Head of the Army AI Centre; David Williams, Chief Technical Officer for the British Army; Brig Richard Byfield MBE, the Army's Chief Digital and Data Officer; Aravindhan Rajasekar, Digital Chief Engineer; and Gary Joynson of the Chief of the General Staff's Action Group. The visit was made possible by members of the cohort: Lt Col Charlie Sprake, Jelena Pecurica and others.

Jelena: How does the Army think about responsible and ethical use of AI, particularly in high-stakes or mission-critical contexts?

Charlie: Our aim is to adopt AI at pace while being grounded in Law, Ethics and Public Trust. This means human-machine teaming is the default approach to AI adoption. We recognise that trust is earned not assumed.

Attendees rated the visit at the highest level available, described the exchange as valuable and insightful and are looking forward to the next one in September.

Signal is written from what members tell us. A few things worth sending in:

• A field visit worth arranging. If you have a connection at an organisation the cohort should see, that is how Andover happened.
• A playbook / guidebook / framework worth passing on. How you approached something and what you learned doing it.
• Something you have published, presented, or written. A paper, a book, a talk.
• Something worth reading. One link and a line on why is enough for the Reading Room.
• An event coming up. Yours or one you think the network should know about.
• What you made of this issue. First issues are drafts in public. Tell us what worked and what did not.
• An idea, or a hand. If you want to help produce this rather than only read it, there is room.
• Anything else. If it is relevant to members, send it and we will find the right place for it.

Write to us with your news, links, and ideas for future issues.

Each issue is our community thinking out loud: what we built, what matters most, and the questions we face as we lead.

© 2026 SIGNAL — Rigor and insight from the Oxford AI Alumni Network.