The Brief with Maya Plentz podcast

Slowing Down AI at the UN

0:00
15:08
Rewind 15 seconds
Fast Forward 15 seconds

Watch my interview with Ian Richards.

The United Nations Secretariat is reinforcing restrictions on the use of generative artificial intelligence, instructing staff to use institutionally approved AI tools rather than independently accessing platforms such as OpenAI’s ChatGPT or Anthropic’s Claude for professional purposes.

Ian Richards, a prominent advocate for UN staff interests, has raised concerns about the restrictions, questioning how the organization expects employees to develop sophisticated AI capabilities while limiting their access to competing technologies.

In a LinkedIn post, Richards reported that IT departments were communicating restrictions covering both installed applications and browser-based access within the UN Secretariat’s working environment. He said Microsoft Copilot remains the authorized option under an organization-wide Microsoft agreement.

The directive is not entirely new. On 5 March 2025, the UN Office of Information and Communications Technology issued updated guidance specifying that Copilot Chat was the only authorized free generative AI chatbot for routine work-related tasks.

The policy concerns professional use within the UN Secretariat, not a universal prohibition across every UN agency, fund and programme, some of which operate under separate technology-governance arrangements.

Richards argues that the restrictions expose a contradiction at the heart of the UN’s modernization agenda.

Through UN 2.0, the Quintet of Change and the UN80 reform initiative, the organization has promoted digital transformation, stronger data capabilities, innovation and artificial intelligence as instruments for improving institutional performance.

Yet staff seeking to experiment with specialized AI platforms can face lengthy approval processes.

In an earlier post, Richards described requirements involving senior management endorsement, review by the Office of Information and Communications Technology, a cross-functional working group on generative AI and potentially the Architecture Review Board.

Such procedures, he argued, risk discouraging employees who are developing applications, automating workflows or using AI for more sophisticated analytical tasks.

The Secretariat has legitimate reasons to impose safeguards. UN personnel handle confidential diplomatic communications, sensitive information about humanitarian beneficiaries, personnel records and material protected by institutional privileges and immunities.

Uploading such information to unauthorized external systems could expose the organization to serious privacy, cybersecurity and legal risks.

But restricting staff to one approved commercial ecosystem raises additional questions about technological dependency, procurement, competition and access to specialized capabilities.

Copilot itself uses large language models, including technology developed by OpenAI. The relevant distinction is therefore not simply which company develops an AI model, but the contractual protections, data-handling arrangements and administrative controls governing its use.

The controversy emerges as the UN seeks to modernize its operations under mounting financial pressure and demands for greater institutional efficiency.

Artificial intelligence could help staff accelerate document analysis, process large datasets, improve translation workflows and develop digital services. Yet these benefits require practical training, secure infrastructure and appropriate access to technology.

Richards warns that overly restrictive policies could prevent staff from pursuing precisely the high-value applications that institutional reform is supposed to encourage.

For the United Nations, the challenge is no longer whether artificial intelligence should be used. It is how to establish credible safeguards without discouraging innovation or concentrating institutional capabilities within a single technology provider.

As the UN promotes global AI governance, its own internal policies are becoming a test of whether it can reconcile technological innovation, institutional accountability and operational security.

Sources: IUnited Nations Office of Information and Communications Technology, Updated UN Guidance on the Use of AI Tools at Work, 5 March 2025.

#TheBrief #UnitedNations #ArtificialIntelligence #UNGeneva #UN80 #AIGovernance #DigitalTransformation #MicrosoftCopilot



This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit theunbrief.substack.com/subscribe

More episodes from "The Brief with Maya Plentz"